Академический Документы
Профессиональный Документы
Культура Документы
TRAINING GUIDE
Server+
Certification
Elton Jernigan, Server+ Certified Technician
03 Serv+ Ch_FM 7/11/01 3:06 PM Page ii
PUBLISHER
SERVER+ CERTIFICATION TRAINING GUIDE David Dwyer
CONTENTS AT A GLANCE
Introduction ............................................................................................................1
A Glossary ..............................................................................................................549
B Overview of the Certification Process ..................................................................561
C What’s on the CD-ROM ....................................................................................563
D Using the ExamGear, Training Guide Edition Software ......................................565
Index ..................................................................................................................587
03 Serv+ Ch_FM 7/11/01 3:06 PM Page iv
03 Serv+ Ch_FM 7/11/01 3:06 PM Page v
Contents
CONTENTS vii
2 1.0—Installation 143
Introduction ......................................................................................................147
Planning for Installation ....................................................................................147
The Installation Plan ....................................................................................147
Verifying the Installation Plan ......................................................................148
Verifying Hardware Compatibility with the Operating System ......................148
Verifying the Installation Site ........................................................................149
Verifying Hardware Components for the Network Server ..............................153
Installing Hardware ............................................................................................161
Electrostatic Discharge ..................................................................................161
Installing Processors ......................................................................................162
Buses ............................................................................................................163
Memory Installation ......................................................................................164
Disk Subsystem Installation ..........................................................................166
Network Subsystem Installation ....................................................................175
Backup System Hardware Installation ............................................................178
Uninterruptible Power-Supply Installation ....................................................178
Miscellaneous Server Components Installation ..............................................179
Redundant Component Installation ..............................................................181
03 Serv+ Ch_FM 7/11/01 3:06 PM Page ix
CONTENTS ix
3 2.0—Configuration 195
Introduction ......................................................................................................198
Network Server Configuration ............................................................................198
Checking the System BIOS ..........................................................................199
Checking the SCSI Controller BIOS ............................................................199
Checking the RAID Controller BIOS ..........................................................200
Configuring Hardware-Based RAID ..................................................................200
Hardware-Based RAID Configuration ..........................................................200
RAID 0 Configuration ..................................................................................203
RAID 1 Configuration (Mirroring) ..............................................................203
RAID 1 Configuration (Duplexing) ..............................................................204
RAID 5 Configuration ................................................................................204
RAID 0/1 Configuration ..............................................................................205
Installing the Network Operating System ..........................................................207
File Systems ..................................................................................................207
Installing the Network Operating System ......................................................213
Configuring the Network ..............................................................................215
Verifying Network Connectivity ....................................................................216
Configuring External Peripherals ........................................................................216
Configuring an External Disk Subsystem ......................................................217
Configuring an External CD-ROM System ..................................................218
Configuring an Uninterruptible
Power Supply ..............................................................................................218
Installing Network Operating System Updates ..................................................220
Configuring Manufacturer-Specific Drivers ........................................................220
Installing Service Tools ......................................................................................221
Installing and Configuring SNMP ................................................................222
Installing Backup Software ............................................................................223
Installing Antivirus Software ..........................................................................223
Installing System Monitor Agents ..................................................................224
03 Serv+ Ch_FM 7/11/01 3:06 PM Page x
4 3.0—Upgrading 239
Introduction ......................................................................................................243
The Upgrade Checklist ......................................................................................243
Locating the Latest Software ..........................................................................244
Reviewing Instructions and FAQs ................................................................244
Testing the Upgrade ......................................................................................245
Scheduling Downtime ..................................................................................245
Performing a Full System Backup ..................................................................245
Implementing Electrostatic Discharge Best Practices ....................................246
Confirming the Upgrade ..............................................................................246
Reviewing the Upgrade and Establishing a New Baseline ..............................247
Documenting the Upgrade ............................................................................247
Performing a Full Backup ..................................................................................247
Adding Processors ..............................................................................................248
Replacing a Single Processor with a Faster Processor ......................................248
Installing Additional Processors ....................................................................249
Upgrading the Operating System for Multiple Processors ..............................253
Adding Hard Drives ..........................................................................................255
Upgrading ATA Hard Disk Drives ................................................................256
Upgrading IDE/ATA/EIDE/ATA-2 Disk Drives to SCSI Disk Drives ..........257
Upgrading SCSI Hard Disk Drives ................................................................258
Adding Drives to a RAID Array ....................................................................259
Adding Memory ................................................................................................260
Check Existing Memory ................................................................................260
Checking Memory Upgrade Feasibility ..........................................................261
Checking Memory Upgrade Compatibility ..................................................262
Installing Additional Memory ........................................................................262
Upgrading the BIOS/Firmware ..........................................................................263
03 Serv+ Ch_FM 7/11/01 3:06 PM Page xi
CONTENTS xi
6 5.0—Environment 321
Introduction ......................................................................................................324
Open-Office Environment ............................................................................324
Server Room ..................................................................................................324
Computer Data Center ..................................................................................325
Environmental Issues ..........................................................................................325
Temperature ..................................................................................................328
Humidity ......................................................................................................331
Air Quality ....................................................................................................332
Power-Line Issues ..........................................................................................332
Fire Suppression ............................................................................................335
Flooding ........................................................................................................336
Electrostatic Discharge ..................................................................................337
Monitoring the Server State ..........................................................................338
Physical Security Issues ......................................................................................339
Physically Securing the Data Center/Computer Room ..................................340
Establish Anti-Theft Devices for Hardware ..................................................344
Limiting Access to Removable Media ............................................................346
Exercises ........................................................................................................350
Review Questions ..........................................................................................351
Exam Questions ............................................................................................351
Answers to Review Questions ........................................................................354
Answers to Exam Questions ..........................................................................355
CONTENTS xiii
A Glossary 549
CONTENTS xv
Index ............................................................................................................587
03 Serv+ Ch_FM 7/11/01 3:06 PM Page xvii
Elton served as a Subject Matter Expert (SME) for the Florida State Enterprise Network Professional
initial focus group for CompTIA's Server+ exam in University: (Windows NT)
June 2000, and as such, helped to define the concepts Enterprise Network Professional
and skills to be covered on the Server+ exam. (Novell NetWare)
Elton believes that everyone should have a hobby and Enterprise Network Professional
his is apparently taking certification exams. Elton holds (Windows 2000)
the following certifications: CNX Certified Network Expert
Microsoft: Microsoft Certified Trainer (MCT) Consortium: (LAN Cabling)
ICCP: Certified Computing Professional Dave Bixler, MCP, MCSE, is the Technical Services
(CCP) Manager and Information Security Officer for a
Prosoft Certified Internet Webmaster Fortune 500 systems integrator. He has been working
Training: (CIW) - Foundations in the computer industry since the mid-1980s, working
on anything from paper tape readers to Windows 2000
Chauncy servers. Dave has lately been focusing on Internet tech-
Group: Certified Technical Trainer (CTT) nologies, specifically computer security, VPNs, and, of
Professsional course, Windows 2000. Dave also has worked on a
Memberships: Network Professional Association number of New Riders books as an author, technical
(NPA) editor, and book reviewer. Dave’s industry certifications
also include Novell’s MCNE and a number of others.
Institute for Network Professionals
Dave lives in Cincinnati, Ohio, with his very patient
(INP)
wife (Sarah), sons (Marty and Nicholas), and two
NetWare Users International (NUI) Keeshonds (Zeus and Arcus).
Building Industry Consulting Service, Emmett Dulaney, MCT, MCSE, CCNA, LPI, and so
International (BICSI) on, is the Director of Training for Mercury Technical
IT Professional Member of the Solutions, the Study Guides Editor for Certification
Computer Technology Industry Magazine, and a monthly contributor to UNIX Review.
Association (CompTIA) He is the author of more than 30 books on various
technologies and certifications. He can be reached at
Elton can be reached at EltonJernigan@hotmail.com.
edulaney@TechnicalContent.com.
ABOUT THE TECHNICAL Jim Cooper, MCSE+I, MCT, has worked in the IT
field for more than 12 years and holds numerous certi-
REVIEWERS fications. He is currently working as a Senior Systems
Engineer deploying an Active Directory and Exchange
These reviewers contributed their considerable hands- 2000 Application Service Provider architecture.
on expertise to the entire development process for Ray Marky, MCP+I, MCSE, MCT, Master Oracle
Server+ Certification Training Guide. As the book was DBA, CCNA, Florida State University Certified
being written, these dedicated professionals reviewed all Webmaster, and CTT has a bachelor of science and a
the material for technical content, organization, and master of science degree in Education from Florida State
flow. Their feedback was critical to ensuring that University. He has taught at the high school and college
Server+ Certification Training Guide fits our reader’s level. He is currently the Director of Technology at the
need for the highest quality technical information. Center for Professional Development at Florida State
University. Ray has also co-authored a book on the
Windows 2000 Server exam.
03 Serv+ Ch_FM 7/11/01 3:06 PM Page xix
DEDICATION xix
Dedication
Elton Jernigan
I dedicate this book to my family, my mother (Elenora),
my wife (Cheryl), and my sons (Daniel and Robby).
Dave Bixler
I want to dedicate this book to my patient wife (Sarah)
and my sons (Marty and Nick). After this is done guys, it’s
out to the swingset!
Acknowledgments
Elton Jernigan
I want to express my deepest appreciation to the follow- Thanks to Dave Bixler for his work on Chapter 6,
ing for their help in getting the Server+ Certification “5.0—Environment,” and Chapter 8, “7.0—Disaster
Training Guide completed. Recovery.”
(Not listed in order of importance!) Thanks to Chris Zahn, New Riders development
editor, for putting up with my endless questions about
Thanks to Jim Vanne, the Server+ guru at CompTIA
the publishing process.
for his hard work getting the Server+ exam from a con-
cept to a reality. Not many people would be interested Thanks to Jay Payne, New Riders media developer
in this book if there were no Server+ exam. for obtaining permissions for all the whitepapers and
software utilities on the CD-ROM.
Thanks to Jeff Riley, New Riders acquisitions editor,
for his constant encouragement and to whom I enjoyed Thanks to my sons (Daniel and Robby) for giving me
proving that, under the correct conditions, a square peg reason not to throw up my hands and quit when the
will indeed fit into a round hole. going got rough.
Thanks to Emmett Dulaney for his work on Chapter 5, Thanks to my wife (Cheryl) for giving me time to work
“4.0—Proactive Maintenance,” and the exam questions on this project.
on the CD-ROM.
03 Serv+ Ch_FM 7/11/01 3:06 PM Page xx
CHAPTER OPENER
Each chapter begins with a set of features
designed to allow you to maximize study
time for that material.
OBJECTIVES
OUTLINE
S T U DY S T R AT E G I E S
Study Strategies: Each topic presents its own
To prepare for the Installation objectives of the . Use the ExamGear test engine on the CD-
Server+ exam, do the following:
learning challenge. To support you through this,
ROM that accompanies this book for addi-
. Read the objectives at the beginning of this tional exam questions concerning this New Riders has included strategies for how to
material.
chapter. best approach studying in order to retain the
. Review the Test Tips available in various places
. Study the information in this chapter.
in this chapter. Make certain that you under-
material in the chapter, particularly as it is
. Review the objectives again. stand the information in the Test Tip. If you addressed on the exam.
. Answer the Review and Exam Questions at the don’t fully understand the Test Tip, review the
end of the chapter and check your results. information in the chapter related to this topic.
03a Serv+ Ch_FM_WT 7/11/01 3:03 PM Page xxiv
INSTRUCTIONAL
EXAM TIP
Know the ESD Best Practices Exam Tip: Exam Tips appear in the
EXAMTIP
Drives Know the STEP BY STEP
IDE/ATA/EIDE/ATA-2 2.1 Installing IDE/ATA/EIDE/ATA-2 Disk Drives
Master/Slave/CSEL requirements.
1. Set the Master/Slave/CSEL jumper on each disk drive. Step by Step: Step by Steps are hands-on tutori-
2. Connect IDE/ATA/EIDE/ATA-2 ribbon cable to al instructions that walk you through a particular
IDE/ATA/EIDE/ATA-2 disk controller. (Red line
on ribbon cable connects to pin 1 on connector on disk task or function relevant to the exam objectives.
controller.)
3. Connect IDE/ATA/EIDE/ATA-2 ribbon cable to
IDE/ATA/EIDE/ATA-2 disk drive. (Red line on ribbon
cable connects to pin 1 on connector on disk drive.)
4. Connect power cable to each IDE/ATA/EIDE/ATA-2
disk drive.
5. Document all settings.
FIGURE 2.27
Rack-mounted keyboard.
IN THE FIELD
In the Field Sidebar: These more extensive
BIG FAN OF DISK MIRRORING
discussions cover material that perhaps is not as
I’m a big fan of disk mirroring. During the useful life (about 4 years)
of the first three file servers that I ever had, one of the mirrored
directly relevant to the exam, but which is useful
disk drives failed in each of the servers. Because the disk drives as reference material or in everyday practice.
were mirrored, none of the users knew that there was a disk drive
failure, and I kept my job. In the Field may also provide useful background
or contextual information necessary for under-
standing the larger topic under consideration.
Key Terms: A list of key terms Chapter Summary: Before the “Apply
appears at the end of each chapter. Your Knowledge” section, you will find
These are terms that you should be a chapter summary that wraps up the
sure to know and are comfortable chapter and reviews what you should
defining and understanding when have learned.
you go in to take the exam.
A P P LY Y O U R K N O W L E D G E
Exercises: These activities provide an opportunity
for you to master specific hands-on tasks. Our
Exercises 3. Does the disk drive use a 50-pin, a 68-pin, or a
80-pin connector?
goal is to increase your proficiency with the prod- 2.1 Becoming Familiar with IDE Disk Drives 4. Attach the cable to the disk drive.
uct or technology. You must be able to conduct You are probably already familiar with IDE drives and
these tasks in order to pass the exam. their installation in a desktop PC. Just in case you are
not, this exercise runs you through some drive basics.
Review Questions
And even if you are familiar with IDE drives, it doesn’t 1. What are the different types of server chassis?
hurt to refresh your memory.
2. How big is one rack unit?
Estimated Time: 10 minutes.
3. How do you convert watts to volt-amps?
1. Get an IDE/ATA or EIDE/ATA-2 disk drive
4. What is the rule on processor steppings when
and cable.
multiple processors are installed in a server?
Review Questions: These open-ended, short- 2. Examine the disk drive. You should notice a
5. What is electrostatic discharge?
answer questions allow you to quickly assess your sticker on the back of the disk drive that gives you
instructions on how to set the disk drive to be a 6. When do you use a UTP cross-over cable?
comprehension of what you just read in the chap- single drive, to be a Master drive, or to be a Slave 7. How many terminators are on a properly termi-
ter. Instead of asking you to choose from a list of drive. You might also see whether the disk drive nated SCSI bus?
options, these questions require you to state the supports CSEL. The jumper settings and their
names are not standardized and therefore differ 8. How many ATA-2 disk drives can be connected
correct answers in your own words. Although you to one ATA-2 channel?
from one manufacturer to another.
will not experience these kinds of questions on
3. Attach the cable to the disk drive. 9. How many Wide SCSI-2 disk drives can be
the exam, these questions will indeed test your connected to one Wide SCSI-2 channel?
level of comprehension of key concepts. 10. What is a KVM switch?
2.2 Becoming Familiar with SCSI
Disk Drives 11. What are the three different types of SCSI bus
The objective of this exercise is the same as that of 2.1, termination?
but as it applies to SCSI drives. You may not have the 12. What is the maximum length of a low-voltage
same level of familiarity with these drives because they differential (LVD) SCSI cable?
are less often encountered in desktop machines.
13. What is the purpose of having redundant power
Estimated Time: 10 minutes. supplies in a network server?
03a Serv+ Ch_FM_WT 7/11/01 3:03 PM Page xxvii
3
4. Servers should be kept away from other devices Answers to Review Questions
that could generate EMI. What does the abbrevi-
ation EMI stand for? 1. A full backup is the most complete backup that
can be made on a system because it includes Answers and Explanations: For each Review and
5. In Windows 2000, the System Properties dialog every file. For more information, see the section
box can be used to identify the processor from “Understanding Backups.”
Exam question, you will find thorough explana-
which tab?
2. Martin is said to be making a baseline. A baseline
tions located at the end of the section.
6. What service must be running in Windows is a gauge of present performance that can be
NT Server 4.0 to be able to receive remote compared against future performance. For more
notifications? information, see the section “Working with
4
Baselines.”
Exam Questions 3. SNMP is the Simple Network Management
Protocol, and it is used to monitor performance
1. Jeff is the network administrator for Mercury on a TCP/IP network. For more information, see
Technical Solutions. He has the server do a full the section “Working with SNMP.”
backup every Sunday and Wednesday night, and
an incremental backup every other night of the 4. EMI is an abbreviation for electromagnetic inter-
week. The server crashed on Wednesday night, in ference. It can be caused by any device using
the middle of the backup. How many tapes does electricity but is common with such things as
he need to do a restore? fluorescent lights, industrial air compressor
5 motors, and the like. For more information, see
A. 1 the section “The Physical Environment.”
B. 2 5. The General tab of the System application will Exam Questions: These questions reflect the
C. 3 show information about the processor and RAM kinds of multiple-choice questions that appear on
in Windows 2000. For more information, see the
D. 4
section “Verifying Hardware.” the Microsoft exams. Use them to become famil-
E. 5 iar with the exam question formats and to help
F. 6 you determine what you know and what you need
G. 7 to review or study more.
Introduction
Server+ Certification Training Guide is designed for those You also will find the features that make New Riders’
with the goal of acheiving certification as a Server+ Training Guides so successful: clear organization, help-
hardware specialist. It covers all the domains and objec- ful hints, tips, real-world examples, and step-by-step
tives identified for the Server+ exam (SKO-001). exercises. Specifically, this book is set up to help you
in the following ways:
According to CompTIA, this exam measures essential
competencies for a server hardware specialist with 18 á Organization. This book is organized according
to 24 months in the server technology area of the com- to individual exam objectives. This book covers
puter industry. You must demonstrate that you have an every objective that you need to know for the
in-depth understanding of planning, installing, config- Server+ exam. The objectives are covered in the
uring, and maintaining servers, including knowledge of same order as they are listed by the certifying
server-level hardware implementations, data storage organization, CompTIA, to make it as easy as
subsystems, data recovery, and I/O subsystems. You possible for you to learn the information. This
should know the interrelationships of all parts of the training guide also attempts to make the informa-
server system and understand the ramifications of their tion accessible in the following ways:
actions. See Appendix B, “Overview of the • This introduction includes the full list of
Certification Process,” for more specifics on what exam topics and objectives.
CompTIA expects in the way of background and
experience for test takers. • Each chapter begins with a list of the objec-
tives to be covered.
This book is your one-stop shop. Everything you need
to know to pass the exams is in here. You do not have • Each chapter also begins with an outline
to take a class in addition to buying this book to pass that provides an overview of the material and
the exam. Depending on your personal study habits or the page numbers of where you can find
learning style, however, you may benefit from buying particular topics.
this book and taking a class. • Objectives are repeated within the text where
This book also can help advanced users and adminis- the material most directly relevant to them is
trators who are not studying for the exam but are covered.
looking for a single-volume technical reference. • Information on where the objectives are
covered is also conveniently condensed on
the tear card located on the CD.
HOW THIS BOOK HELPS YOU á Instructional features. This book has been
designed to provide you with multiple ways to
This book provides a self-guided tour of all the areas learn and reinforce the exam material. Following
covered by the Server+ exam and identifies the specific are some of the helpful methods:
skills you need to achieve your Server+ certification.
04 Serv+ Ch_Intro 7/11/01 3:14 PM Page 2
• Objective explanations. As mentioned previously, • Step by Steps. These are hands-on, tutorial
each chapter begins with a list of the objectives instructions that lead you through a particular
covered in the chapter. In addition, immediate- task or function relevant to the exam objectives.
ly following each objective is an explanation in
• Exercises. Exercises are provided at the ends of
a context that defines it more meaningfully.
chapters, providing you with more opportu-
• Study strategies. The beginning of each chapter nities for hands-on learning and reinforce-
also includes strategies for studying and ment of the concepts.
retaining the material in the chapter, particu-
á Extensive practice test options. This book pro-
larly as it is addressed on the exam.
vides numerous opportunities for you to assess
• Exam tips. Exam tips appear in the margin to your knowledge and to practice for the exam.
provide specific exam-related advice. Such tips The practice options include the following:
may address what material is covered (or not
• Review Questions. These open-ended ques-
covered) on the exam, how it is covered,
tions appear in the “Apply Your Knowledge”
mnemonic devices, and particular quirks of
section at the end of each chapter. They
that exam.
enable you to quickly assess your comprehen-
• Review breaks and summaries. Crucial infor- sion of what you just read in the chapter.
mation is summarized at various points in the Answers to the questions are provided later in
book in lists or tables. Each chapter ends with the section.
a summary as well.
• Exam Questions. These questions also appear
• Key terms. A list of key terms appears at the in the “Apply your Knowledge” section. They
end of each chapter. reflect the kinds of multiple-choice questions
that appear on the Server+ exam. Use them to
• Notes. These appear in the margin and contain
practice for the exam and to help you deter-
various kinds of useful information such as tips
mine what you know and what you need to
on technology or administrative practices, his-
review or study further. Answers and explana-
torical background on terms and technologies,
tions for them are provided.
or side commentary on industry issues.
• Practice exam. A practice exam is included in
• Warnings. When using sophisticated tech-
the “Final Review” section for each exam (as
nology improperly, the potential for mistakes
discussed later).
or even catastrophes to occur is ever present.
Warnings appear in the margin to alert you to • ExamGear. The ExamGear software included
such potential problems. on the CD that accompanies this book pro-
vides even more practice questions.
• In the Fields. These more extensive discussions
cover material that may not be directly related
to the exam but which is useful as reference
NOTE
material or in everyday practice. In the Fields More About ExamGear For a com-
also provide useful background or contextual plete description of the New Riders
information necessary for understanding the ExamGear test engine, see Appendix
larger topic under consideration. D, “Using the ExamGear, Training
Guide Edition Software.”
04 Serv+ Ch_Intro 7/11/01 3:14 PM Page 3
INTRODUCTION 3
NOTE
following three valuable tools that can help you Post-Beta Objectives The objectives
prepare for the exam: detailed in this section cover the
Server+ exam blueprint. This docu-
ment was produced after the final
• Fast Facts. This condensed version of the
technical and psychometric review of
information contained in the book will prove
the item pool following the beta-test-
extremely useful for last-minute review. ing period. This document reflects the
• Study and Exam Prep Tips. Read this section topics and technologies that appear
early on to help you develop study strategies. as part of the Server+ exam.
It also provides valuable exam-day tips and The objectives listed here represent
information. CompTIA’s outline of February 21,
2001. This certification exam went
• Practice Exam. A full practice test is included.
live on February 14, 2001.
Questions are written in the styles used on
the actual Server+ exam. Use it to assess your
readiness for the real thing. The skills and knowledge measured by the examination
are derived from an industry-wide job-task analysis and
This book includes several valuable appendixes as well, validated through a worldwide survey of more than 400
including a glossary (Appendix A), an overview of the professionals, each of whom had a minimum of 18
Server+ certification program (Appendix B), a descrip- months of experience with advanced PC hardware. The
tion of what is on the CD-ROM (Appendix C), and an results of the worldwide survey were used in weighting
explanation of the ExamGear test engine (Appendix D). the domains and ensuring that the weighting is represen-
These and all the other book features mentioned tative of the relative importance of that content to the
previously will enable you to thoroughly prepare for job requirements of server hardware specialist with 18 to
the exam. 24 months of experience. You can find the results of the
job-task analysis and survey in the following report:
For information on how to register for the exam see
Appendix B. á Server+ Job Task Analysis (JTA) (August 2000)
INTRODUCTION 5
2.4 Configure external peripherals (UPS, external 3.3 Add hard drives
drive subsystems, etc.)
• Verify that drives are the appropriate type
2.5 Install NOS updates to design specifications
• Confirm termination and cabling
2.6 Update manufacturer specific drives
• For ATA/IDE drives, confirm cabling,
2.7 Install service tools (SNMP, backup software, master/slave and potential cross-brand
system monitoring agents, event logs, etc.) compatibility
2.8 Perform server baseline • Upgrade mass storage
2.9 Document the configuration • Add drives to array
• Replace existing drives
• Verify that server and OS recognize the added 3.8 Upgrade system monitoring agents
memory
• Perform upgrade checklist, including: locate
• Perform server optimization to make use of and obtain latest test drivers, OS updates,
additional RAM software, etc.; review FAQs, instructions,
facts and issues; test and pilot; schedule
3.5 Upgrade BIOS/firmware
downtime; implement using ESD best prac-
• Perform upgrade checklist, including: locate tices; confirm that the upgrade has been
and obtain latest test drivers, OS updates, recognized; review and baseline; document
software, etc.; review FAQs, instructions, facts the upgrade
and issues; test and pilot; schedule downtime;
implement using ESD best practices; confirm 3.9 Upgrade service tools (for example, diagnostic
that the upgrade has been recognized; review tools, EISA configuration, diagnostic partition,
and baseline; document the upgrade SSU, etc.)
• Perform upgrade checklist, including: locate
3.6 Upgrade adapters (e.g., NICs, SCSI cards,
and obtain latest test drivers, OS updates,
RAID, etc.)
software, etc.; review FAQs, instructions,
• Perform upgrade checklist, including: locate facts and issues; test and pilot; schedule
and obtain latest test drivers, OS updates, downtime; implement using ESD best prac-
software, etc.; review FAQs, instructions, facts tices; confirm that the upgrade has been
and issues; test and pilot; schedule downtime; recognized; review and baseline; document
implement using ESD best practices; confirm the upgrade
that the upgrade has been recognized; review
and baseline; document the upgrade 3.10 Upgrade UPS
• Perform upgrade checklist, including: locate
3.7 Upgrade peripheral devices, internal and external
and obtain latest test drivers, OS updates,
• Perform upgrade checklist, including: locate software, etc.; review FAQs, instructions,
and obtain latest test drivers, OS updates, facts and issues; test and pilot; schedule
software, etc.; review FAQs, instructions, downtime; implement using ESD best prac-
facts and issues; test and pilot; schedule tices; confirm that the upgrade has been
downtime; implement using ESD best prac- recognized; review and baseline; document
tices; confirm that the upgrade has been the upgrade
recognized; review and baseline; document
the upgrade
04 Serv+ Ch_Intro 7/11/01 3:14 PM Page 7
INTRODUCTION 7
• Locate and effectively use hot tips (e.g., fixes, • Develop disaster recovery plan
OS updates, E-support, web pages, CDs)
• Identify types of backup hardware
• Gather resources to get problem solved: • Identify types of backup and restoration
schemes
• Identify situations requiring call for
• Confirm and use off site storage for backup
assistance
• Document and test disaster recovery plan reg-
• Acquire appropriate documentation ularly, and update as needed
• Describe how to perform remote trou- • Identify hot and cold sites
bleshooting for a remote alert • Implement disaster recovery plan
INTRODUCTION 9
NOTE
Exam-Taking Advice Although this
book is designed to prepare you to
More extensive tips are found in the “Final Review” take and pass the Server+ exam,
section titled “Study and Exam Prep Tips,” but keep there are no guarantees. Read this
this advice in mind as you study: book, work through the questions and
exercises, and when you feel confi-
á Read all the material. Make sure your exam
dent, take the practice exam and addi-
preparation is thorough. Do not just drop into
tional exams using the ExamGear test
the book and read around. Read through all the engine. This should tell you whether
material. This book has included additional you are ready for the real thing.
information not reflected in the objectives in an
When taking the actual certification
effort to give you the best possible preparation
exam, make sure you answer all the
for the examination—and for the on-the-job
questions before your time limit
experiences to come.
expires. Do not spend too much time
á Do the Step by Steps and complete the on any one question. If you are
exercises. This will provide you with another way unsure, answer it as best as you can;
of understanding the material, as well as more then mark it for review after you have
finished the rest of the questions.
information on how well you comprehend it.
á Use the questions to assess your knowledge.
Do not just read the chapter content; use the
questions to find out what you know and what
you do not know. Study some more, review, and
then assess your knowledge again.
á Review the exam objectives. Develop your own
questions and examples for each topic listed. If
you can develop and answer several questions for
each topic, you should not find it difficult to pass
the exam.
Remember, the primary object is not to pass the
exam—it is to understand the material. After you
understand the material, passing the exam should be
simple. Knowledge is a pyramid; to build upward, you
need a solid foundation. This book and the CompTIA
Server+ certification program are designed to ensure
that you have that solid foundation. For information
on how to schedule an exam, see Appendix B, or
visit www.comptia.org or www.comptia.com.
Good luck!
04 Serv+ Ch_Intro 7/11/01 3:14 PM Page 10
05 Serv+ Ch_P1 7/10/01 9:40 AM Page 11
I
P A R T
EXAM PREPARATION
2 1.0—Installation
3 2.0—Configuration
4 3.0—Upgrading
5 4.0—Proactive Maintenance
6 5.0—Environment
8 7.0—Disaster Recovery
05 Serv+ Ch_P1 7/10/01 9:40 AM Page 12
06 Serv+ Ch_01 7/11/01 9:59 AM Page 13
OBJECTIVES
0.0—Planning and
Specifications
06 Serv+ Ch_01 7/10/01 9:41 AM Page 14
OBJECTIVES
OBJECTIVES
0.4 Identify the major Intel processors and . The server hardware specialist should be able to
the importance of clock speed, processor determine which memory configuration will best
cache speed and size, and symmetric allow the network server to meet the needs of its
multiprocessing (SMP): assigned role in the network.
• Intel processors 0.7 Identify the strengths and weaknesses of
the various types of disk subsystems.
• Processor cache
Content may include the following:
• Symmetric multiprocessing (SMP)
• Integrated Drive Electronics (IDE/ATA)
. The server hardware specialist must determine the
• Enhanced Integrated Drive Electronics
appropriate Intel processor(s) to select based on the
category of the network server and the role of the (EIDE/ATA2)
network server in the network. The selection of the • Ultra ATA / Ultra DMA
proper processor(s) will dictate the processor clock
speed, the processor cache speed and size, and • Serial ATA
whether symmetric multiprocessing can be sup- • Small Computer System Interface (SCSI)
ported by the network server.
• Self-Monitoring, Analysis and Reporting
0.5 Identify the various buses that are part
Technology (S.M.A.R.T.)
of the network server. The major buses
include: • Fibre Channel
• Memory bus • Fibre Channel Arbitrated Loop (FC-AL)
• System bus • Serial Storage Architecture (SSA)
. The speed of the memory bus determines how fast • Storage area network (SAN)
data can be moved from the processor to memory
and from the memory to the processor. The type, • storage (NAS)
speed, width, and number of the system buses . The server hardware specialist must be able to select
determines how quickly data can move from out- the best disk subsystem that will meet the role that
side the network server to inside the network server. the network server will perform on the network.
0.6 Identify the characteristics of different 0.8 Identify different network subsystems.
types of memory. Identify how the amount Successfully attach various devices to
of memory affects network server perfor- and remove various devices from the net-
mance. Identify how interleaving of mem- work subsystem. Content may include the
ory affects network server performance. following:
Major memory topics include:
• Ethernet
• Amount of memory
• Token Ring
• Memory interleaving
• Network interface cards (NICs)
• Type of memory
06 Serv+ Ch_01 7/10/01 9:41 AM Page 16
OBJECTIVES
. Server hardware specialists should determine the . Server hardware specialists should be able to
degree of redundancy required by the network server describe and facilitate the implementation of
and which hardware components need redundancy. backup and recovery strategies that meet the needs
of the role of the network server.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 17
OBJECTIVES
0.15 Identify various methods to provide the 0.18 Identify situations when network servers
network server with fault tolerance. should be installed in a server rack and
Fault-tolerance methods include: the proper installation of the network
server into a server rack. Identify special-
• Uninterruptible power supplies ized components and installation require-
• Redundant arrays of inexpensive disks ments for rack-mounted network servers.
(RAID) Content may include the following:
. Server hardware specialists should understand the • EIA rack unit (1.75 inches)
requirement of providing an uninterruptible power • Keyboard/video/monitor (KVM) switch
supply for a network server. The redundancy,
clustering, scalability, and high availability sections • Cable management
cover other fault-tolerant topics.
• Rack layout
0.16 Identify techniques and plan for recovery
from a disaster (total failure of the • Security
network server). • Cooling
. Server hardware specialists should be able to plan • Rack-mounted keyboard/trackball
for recovery from a disaster (total failure or loss of
the network server). The final product from this • Rack-mounted monitor/LCD panel
planning effort is a disaster-recovery plan.
• Rack-mounted uninterruptible power
0.17 Identify miscellaneous hardware compo- supply (UPS)
nents that are part of a network server.
Content may include the following:
. Server hardware specialists should be able to iden-
tify when a network server should be mounted into
• Video controller a server rack. A server hardware specialist should be
able to select and install the specialized equipment
• Video monitor associated with rack-mounted network servers.
• Universal Serial Bus 0.19 Identify network operating systems. A
• FireWire server hardware specialist should be
able to recognize the characteristics of
• Keyboard various network operating systems.
Content may include the following
• Mouse
network operating systems:
• CD-ROM drive
• Novell NetWare 4
• DVD-ROM drive
• Novell NetWare 5
• Floppy disk drive
• Microsoft Windows NT Server
• Modem
• Microsoft Windows 2000 Server
. Server hardware specialists should be able to select
the proper miscellaneous hardware that is part of a • UNIX
functioning network server.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 18
OBJECTIVES OUTLINE
• Linux Introduction 23
• IBM OS/2 Warp Server
. Server hardware specialists should be able to Network Types 23
identify various network operating systems. A
server hardware specialist should be able to install, Peer-to-Peer Networks 23
start up, shut down, and perform simple trouble- Server-Centric Networks 24
shooting on the major network operating systems.
0.20 Identify hardware and software available Network Server Categories 24
to manage network servers. Content
may include the following: Workgroup Server 25
• Vendor-specific installation and configuration Departmental Server 25
software Enterprise Server 26
Server Appliances 26
• Vendor-specific management software
• Vendor-specific management hardware
Network Server Roles 26
• Antivirus software
Gateway Server 28
• Documentation
Router 28
. Server hardware specialists should be able to Bridge 29
identify and use various hardware and software
Firewall 30
tools to manage a network server.
Network Address Translation Server 30
0.21 Creating an installation plan
Proxy Server 31
. Server hardware specialists should understand
Database Server 32
the process of creating an installation plan. The
first step in installing a new network server is to Client/Server 32
verify the installation plan. After verification, the Application Server 32
installation plan is used as a guide for the network Mail Server 33
server installation.
FTP Server 34
SNA Server 34
Remote Access Server 35
Remote Control Remote Access
Technique 35
Remote Node Remote Access
Technique 35
06 Serv+ Ch_01 7/10/01 9:41 AM Page 19
OUTLINE
OUTLINE
OUTLINE
S T U DY S T R AT E G I E S
To successfully learn the material outlined in the . Review the Exam Tips available in various places
objectives listed in this chapter, do the following: in this chapter. Make certain that you under-
stand the information in the Exam Tip. If you
. Read the objectives at the beginning of this
don’t fully understand the Exam Tip, review the
chapter.
information in the chapter related to this topic.
. Study the information in this chapter.
. Due to the length of this chapter, it contains
. Review the objectives again. several Review Breaks, which summarize the
information contained in the immediately pre-
. Answer the Review and Exam Questions at the
ceding section of the chapter. In most cases,
end of the chapter and check your results.
the Review Breaks contain recommendations
and suggestions on network server technology.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 23
INTRODUCTION
The goal of this chapter is to introduce all the concepts related to
network servers. The components that can be used to configure a
network server so that it can effectively perform its designated role
in the network are discussed in detail. The information in this
chapter should help the network engineer select the appropriate
components for the network server. The end goal is the creation of
an installation plan that can be used by a server hardware specialist
to install the network server to the specifications created by the
network engineer.
NETWORK TYPES
. 0.1 Identify the two types of networks
• Peer-to-peer network
• Server-centric network
Networks are of two types: peer-to-peer and server-centric (server-
based), which also is sometimes called a client/server network.
Peer-to-Peer Networks
NOTE
Server-Centric Networks
NOTE
• Enterprise server
• Server appliances
Network servers come in various sizes or categories. They range from
the smallest “glorified desktop” computer to the enterprise-sized
mainframe, such as the Unisys ES7000, which can have up to 32
processors, 64GB of memory, and 96 PCI bus slots. The Unisys
ES7000 is in the one million dollar price range. For more information
on the Unisys ES7000, visit www.unisys.com/hw/servers/es7000/.
Most network servers are configured and priced somewhere in
between these two extremes.
Workgroup Server
A workgroup server is a network server that can support a limited
number of client computers (generally fewer than 20) or a network
service that doesn’t require many resources, such as a DHCP or
DNS server. The term “workgroup” is very fitting for this category
of network server. A workgroup server generally has a single proces-
sor, a relatively small amount of memory, and limited disk storage
capacity. A workgroup network server has limited scalability. (It
cannot add multiple processors, the maximum amount of memory
is relatively small, and the space for internal disk drives is very
limited.) A workgroup network server is sometimes called a Small
Office Home Office (SOHO) server.
Departmental Server
A departmental server is a network server that can support up to
approximately 300 client computers. A departmental server might
also provide a network service that requires considerable resources,
such as a web server. A departmental server can have multiple
processors, a considerable amount of memory, and a large amount of
disk storage. A departmental server is generally scalable to a large
network server with more processors, a great deal of additional
memory, and additional disk storage.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 26
Enterprise Server
An enterprise server is an extremely large server designed to support
thousands of clients. It might also provide a network service, such as
a database server, to a very large number of users. Enterprise servers
nearly always have multiple processors, a tremendous amount of
memory, and a gigantic amount of disk storage. Enterprise network
servers are extremely scalable. They could have as many as 32 proces-
sors, as much as 64GB of memory, and terabytes of disk storage.
Server Appliances
A server appliance is a relatively new category of network server. A
server appliance is a network server designed to provide a single ser-
vice to the network and to take a minimum amount of space in the
server room. Many times these server appliances are designed to be
mounted in a server rack and take no more than 1.75 inches (1 rack
unit) of vertical space.
A good example of a server appliance would be the use of one of these
physically small servers configured to function as a web server. Other
good examples of a server appliance would be a physically small
network server configured to function as a DNS server or a firewall.
• Application server
• Mail server
• FTP server
• SNA server
• Remote access server (RAS)
• File and print server
• Terminal server
• Fax server
• Domain Name Server (DNS)
• Windows Internet Naming Server (WINS)
• Dynamic Host Configuration Protocol Server (DHCP)
• Web server
• Video server
• CD-ROM library server
• Directory services server
A network server can assume different jobs or roles in the network.
It is also common for a single network server to perform multiple Network Server Roles Know the
TIP
roles simultaneously (such as DNS, DHCP, and WINS). Because of basics of the different roles that a
the resources required, however, some network server roles practi- network server can perform in a
EXAM
network.
cally require that the network server perform a single role, such as a
database server.
Installing a service or application defines the role that a network
server will perform. For example, installing the DHCP service on a
network server gives it the DHCP server role. This server will often
be referred to as the DHCP server. Installing a database application,
such as Oracle or Microsoft SQL Server, defines the network server
as a database server.
A network server running multiple services, such as DHCP, DNS,
and WINS, might be called the DHCP server, the DNS server, or
the WINS server, depending on who is referring to it or what is the
current topic of discussion. The fact that a single network server is
actually performing all three roles simultaneously isn’t necessarily
important to the conversation.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 28
The major roles that a network server might perform are discussed
in the following sections.
Gateway Server
A gateway is a network device that connects two totally dissimilar
communications systems, such as a microcomputer network and a
mainframe.
The classic example of a gateway server is supplying communica-
tions between a local area network (LAN) and an IBM mainframe.
The LAN could be using Ethernet and the TCP/IP protocol,
whereas the IBM mainframe communicates could be using IBM’s
System Network Architecture (SNA) protocol. The gateway server is
connected to both the LAN and the SNA communications system
and translates the protocols and data between the two systems.
The following software products allow a network server to function
as a gateway server:
á Microsoft SNA Server
á Microsoft Host Integration Server
á Novell NetWare for SAA
Router
A router is a network device that moves data packets from one net-
work (as defined by the data packet’s source network address) to
another network (as defined by the data packet’s destination network
address). A network server can perform the role of a router just by
having multiple network interface cards (NICs), with each NIC
attached to a different network. It is a simple operation to configure
the network operating system to send data between the NICs in the
network server.
Most network operating systems can function as a router by enabling
packet forwarding between multiple NICs. Configuration of packet
forwarding is specific to each different network operating system.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 29
Bridge
A bridge is a device that “filters” frames based on the hardware
address of the device that sent the frame and the hardware address
of the device that is to receive the frame. This is useful in the case
where you want to isolate data packets on one segment of the
network from another segment of the network.
For example, you could use a bridge to connect a network in a
building to a backbone network that connects multiple buildings
together. If the network packet being sent is destined to a computer
within the building, there is no need for it to be forwarded to the
backbone, so the bridge does not forward the packet to the back-
bone. If the network packet is being sent to a computer in another
building, however, the bridge must forward the packet to the back-
bone segment for it to reach its destination.
A network server can be configured to perform the function of a
bridge. The network server will need to have two NICs installed and
be configured to forward data packets based on the hardware address
of the sending and receiving computers.
The most important component of a network server that is going to
function as a bridge is the network subsystem.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 30
Firewall
A firewall is a network device that forwards data packets from one
network to another network based on rules defined by the network
administrator. These rules can reject or accept data packets based on
many different criteria, including:
á Destination address
á Source address
á Network service based on TCP/IP port address
A firewall examines each and every data packet and decides whether
the packet will be forwarded to another network based on the rules
defined by the administrator who configured the firewall. Generally,
firewalls connect a corporate LAN to the Internet. The firewall is
configured to prevent access from the Internet to the corporate
LAN. The firewall can also restrict access from the corporate LAN to
the Internet. Examples of firewall software are as follows:
á PGP Security’s Gauntlet
á Check Point’s Firewall-1
á CyberGuard’s Firewall
á Microsoft Internet Security and Acceleration (ISA) Server
valid IP addresses that are used by client data packets destined for
the Internet. The NAT server actually changes the private IP address
to a valid IP address as the client’s data packet goes from the private
network to the Internet. When a reply to the data packet that was
sent is received, the NAT server changes the valid IP address back to
the private IP address of the client on the private network.
The most important components of a network server that is going to
function as a NAT server are the network subsystem and the processor.
Microsoft Windows 2000 Server includes NAT capability as part of
the Routing and Remote Access Service (RRAS).
Microsoft Internet Security and Acceleration (ISA) server extends
Windows 2000 NAT capability by providing SecureNAT to increase
NAT security.
Proxy Server
A proxy server allows a client computer’s web browser access to the
World Wide Web through a single access point. The proxy server is
NOTE
considered by some people to be a firewall; but as a firewall, a proxy Private IP Addresses As defined by
server’s capabilities are rather limited. The big advantage that a RFC 1918, “Address Allocation for
proxy server has is its capability to cache frequently accessed web Private Internets,” the reserved
addresses to be used for private net-
pages on its own hard disk drives. Having web pages cached locally
works are allocated in three ranges:
reduces the frequency of access to the Internet by the proxy server.
Using a proxy server requires only that there be one valid IP address • 1 Class A address from 10.0.0.0
assigned to the corporation. That address is assigned to the proxy to 10.255.255.255
server. All other client computers connect to web sites on the • 16 Class B addresses from
Internet using the proxy server. 172.16.0.0 to 172.31.255.255
Examples of proxy servers include the following: • 255 Class C addresses from
192.168.0.0 to 192.168.255.255
á Microsoft Proxy Server
á Microsoft Internet Security and Access Server
á Ositis Software’s WinProxy
á Netscape’s Proxy Server
Database Server
A database server is the classic example of a client/server system. A
database server runs a database management software package. The
database management software allows clients to create, input,
update, and retrieve data. All the data resides on the disk drives
attached to the database server. All processing of the data occurs on
the database server.
A client issues a request or query for data. The database server
processes the data in the database(s) on its disk drives and returns
the results to the client.
The three most important components of a network server that is
going to function as a database server are processors, memory, and
the disk subsystem.
Examples of database software include the following:
á Microsoft SQL Server 2000
á Oracle
á Sybase
Client/Server
“Client/server” is a generic term that describes the relationship
between computers on the network. A client computer is a computer
that requests the use of resources on other computers on the net-
work. A server computer is a computer that shares its resources with
other computers on the network. The resources can be physical,
such as files on disk drives or a printer, or the resources can be
services provided by the server.
Application Server
An application server is a network server that runs a network
application. There are three models for applications servers:
á Dedicated application model. This is the most common
application server model. The application server runs a net-
work application, such as a database, that can be accessed by
clients on the network. The dedicated application server does
not provide any other service to the network.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 33
Mail Server
A mail server, or more accurately an email server, provides messaging
services to a network. The email server usually uses the TCP/IP
Simple Mail Transfer Protocol (SMTP) to deliver email messages
between servers either on a LAN or the Internet. Client computers
often use email client protocols, such as POP (Post Office Protocol)
or IMAP (Internet Message Access Protocol), to retrieve email mes-
sages from SMTP email servers.
Examples of software that allows a network server to function as an
email server include the following:
á Microsoft Exchange Server
á Novell GroupWise
á Sendmail for UNIX/Linux
FTP Server
An FTP server uses the TCP/IP File Transfer Protocol (FTP) to allow
access to files over the network. Before the advent of web servers,
FTP servers provided access to files over the Internet. To access files
on an FTP server, the client computer must use an FTP client or a
web browser that understands the FTP protocol. Most client com-
puter operating systems have a simple command-line-driven FTP
client available as a standard part of the operating system. Graphical
versions of the FTP client software, which are much easier to use
than the command-line version, are available from third parties.
For Microsoft Windows NT Server and Microsoft Windows 2000
Server, the FTP service software is part of Microsoft Internet
Information Server (IIS), which is a standard part of the
operating system. FTP service for UNIX systems is provided
by the FTP daemon.
The most important component of a network server that is going to
serve as an FTP server is the network subsystem. If a large number
of files are going to be on the FTP server, the amount of disk storage
is also important.
SNA Server
NOTE
Terminal Server
A terminal server supports clients by running applications that the
clients cannot run themselves because of lack of resources, such as
processor speed or amount of memory. These clients are often
referred to as “thin” clients.
A client establishes a session with the terminal server and runs stan-
dard desktop applications, such as a word processor or an electronic
spreadsheet, on the terminal server itself.
For the terminal server to perform its function, it needs a tremen-
dous amount of resources. Multiple fast processors are required. A
great amount of memory is also required because each client can use
10 to 20MB of memory on the terminal server to run applications.
A terminal server can be created using Microsoft Windows NT
Server 4 Terminal Services Edition, or Microsoft Windows 2000
Terminal Services, which is a standard part of Windows 2000 Server,
Windows 2000 Advanced Server, and Windows 2000 Datacenter
Server. Another popular solution used to implement a terminal
server is Citrix Metaframe. For more information about Citrix
Metaframe, visit their web site at www.citrix.com/.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 37
Fax Server
A fax server adds the capability to send facsimiles to and receive
facsimiles from the network. A fax server can also route incoming
faxes directly to users’ computers, share fax lines for outgoing faxes,
and even send faxes by way of the Internet.
Example software products that allow a network server to function
as a fax server include the following:
á Castelle FaxPress
á Cheyenne FaxServe
Web Server
A web server provides information content over the network to client
computers using a web browser, such as Microsoft Internet Explorer
or Netscape Navigator by using TCP/IP Hypertext Transfer Protocol
(HTTP). The information content is created on the web server
using the Hypertext Markup Language (HTML).
Examples of software that allow a network server to function as a
Web server include the following:
á Microsoft Internet Information Service (IIS)
á Apache Web Server
á Netscape Enterprise Web Server
Video Server
A video server stores digitized video on its disk drives for replay
through the network.
The important components for a network server that is going to
function as a video server are the disk subsystem and the network
subsystem. A video capture card is also required to convert analog
video from a VCR or video camera into digital video to be stored on
the video server’s disk drives.
A video server can be used to create on-demand training, web
broadcasts using streaming media, and product demonstrations over
the web.
An example of software that can be used to create a video server is
Microsoft Windows Media Technologies.
This section described the various roles that a network server can R E V I E W B R E A K
perform in a network. A network server role is defined by the ser-
vices or applications that it provides to the network. The four major
network server components—processor, memory, disk subsystem,
and network subsystem—are of different importance depending on
the network server role.
When making the decision on the configuration of a new network
server, the role that the network server will play in the network as
well as the number of clients that the network server will service
are of paramount importance. Undersizing the network server will
lead to many unsatisfied clients. Oversizing the server is a waste of
valuable resources.
INTEL PROCESSORS
. 0.4 Identify the major Intel processors and the importance of
clock speed, processor cache speed and size, and symmetric
multiprocessing (SMP):
• Intel processors
• Processor cache
• Symmetric multiprocessing (SMP)
The CompTIA Server+ exam deals only with network servers that
are based on processors built by Intel Corporation. This section
discusses the various Intel processors that you might encounter in
a network server and their characteristics.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 42
Intel Processors
NOTE
Pentium Pro
The Pentium Pro processor was built specifically for network servers.
It had a Level 2 cache that operated at the same speed as the proces-
sor itself. It was also designed for multiprocessing. Many network
servers were built with two or four Pentium Pro processors inside.
Specifications for the Pentium Pro include the following:
á Date of service (November 1995)
á Speeds 150, 166, 180, 200MHz
á Level 2 cache 256KB, 512KB, 1024KB full core speed
Pentium II
The Pentium II processor had a faster processor clock than the
Pentium Pro, but the Level 2 cache ran at one-half the speed of the
processor. Multiprocessing with Pentium II processors was limited to
a maximum of two processors. Specifications for the Pentium II
include the following:
á Date of service (May 1997)
á Speeds 233, 266, 300, 333, 350, 400, 450MHz
á Level 2 cache 256KB, 512KB running at one-half core speed
Pentium II Xeon
The Pentium II Xeon processor was designed for network servers. It
had a faster processor speed than the Pentium Pro, its Level 2 cache ran
at full processor speed, it had larger Level 2 caches available, and multi-
processing network servers could be configured with up to four proces-
sors. Specifications for the Pentium II Xeon include the following:
á Date of Service (June 1998)
06 Serv+ Ch_01 7/10/01 9:41 AM Page 43
Pentium III
The Pentium III processor offered faster processor speeds than previ-
ous processors and in some cases the Level 2 cache ran at the speed
of the processor. However, multiprocessing was limited to two
processors. Specifications for the Pentium III include the following:
á Date of service (February 1999)
á Speeds 450, 500, 533, 550, 600, 650, 667, 700, 733, 750,
800, 850, 866, 933, 1000, 1130MHz
á Level 2 cache 256KB, 512KB one-half or full core speed
Pentium 4
The Pentium 4 processor is not well suited for network servers.
Although it has fast processor speeds, its Level 2 cache is rather
small, and there is no multiprocessing option available. The Pentium
4 processor is designed for advanced workstations, not network
servers. Specifications for the Pentium 4 include the following:
á Date of service (November 2000)
06 Serv+ Ch_01 7/10/01 9:41 AM Page 44
Itanium
All the Pentium processors are 32-bit processors. The Intel Itanium
is Intel’s first 64-bit processor. The Itanium is backward compatible
with existing Intel 32-bit processors without software emulation,
meaning that an existing 32-bit program will run without modifica-
tion on the Itanium processor. Initially, up to 4 Itanium processors
can be installed in a network server to support SMP environments.
Specifications for the Itanium include the following:
á Date of service (not released at the time of this writing)
á Speeds 733MHz, 800MHz
á Level 2 cache 96KB
á Level 3 cache 2MB or 4MB
A good white paper that details the benefits of the Itanium processor
can be found at ftp://download.intel.com/design/
IA-64/Downloads/ia64rev2.pdf.
Processor Cache
Processor cache is high-speed static random access memory (SRAM)
rather than the slower and cheaper dynamic random access memory
(DRAM) used for system memory. Processor caching is used because
most programs access the same data and instructions multiple times.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 45
Symmetric Multiprocessing
NOTE
R E V I E W B R E A K
This section has general information about several different Intel
processors. When selecting the processor for a network server, and
considering the role that the network server will perform in the net-
work, pick the fastest processor with the most Level 2 cache that fits
the budget of the network server. To provide for future scalability of
the network server, select a network server that can support SMP.
SERVER BUSES
. 0.5 Identify the various buses that are part of the network
server. The major buses include:
• Memory bus
• System bus
There are many buses in a network server. Buses are a mechanism for
moving data, addresses, and control information between compo-
nents in the network server. Of particular importance are the mem-
ory bus and the system bus. Bus speeds are measured in megahertz
(millions of cycles per second). The faster the bus speed, the more
work that can be done in a given amount of time. Buses cannot be
upgraded without replacing the entire system board in the network
server. Therefore, get the fastest bus that is available and that fits the
budget for the network server.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 47
Memory Bus
NOTE
Bus Transfer Rates The speed of
The memory bus connects the main memory with the processor of buses is measured in megahertz
the network server. This is often referred to as a frontside bus. There (MHz). Because the buses transfer
is also a backside bus that connects the processor with its Level 2 multiple bits in parallel for each cycle
cache. The faster the frontside bus, the faster the processor can (hertz), the transfer rates are speci-
access memory. Currently, frontside buses are available that operate fied in megabytes per second. A byte
at 66MHz, 100MHz, and 133MHz. is generally 8 bits, and bus widths are
designed in multiples of 8 bits. An
Local buses are an extension of the frontside bus. 8-bit bus transfers 1 byte per cycle. A
16-bit bus transfers 2 bytes per cycle.
A 32-bit bus transfers 4 bytes per
System Bus cycle. A 64-bit bus transfers 8 bytes
per cycle.
There have been many system buses through the history of the
microcomputer. Table 1.1 is a comparison of the various systems
buses. Following the table are brief descriptions of the different buses.
TABLE 1.1
S Y S T E M B U S C O M PA R I S O N
Bus, Bus Width Bus Speed (MHz) Transfer Speed
Year Introduced (Bits) Megabytes/Sec.
ISA (8-bit), 1981 8 4.77 4.77
ISA (16-bit), 1984 16 8 16
EISA, 1988 32 8 32
Micro Channel, 1987 16 8 16
Micro Channel, 1987 32 8 32
VESA Local Bus, 1992 32 33 132
PCI, 1992 32 33 132
PCI, 1992 64 33 264
PCI 2.1, 1995 64 66 528
PCI-X 1.0, 1999 64 132 1056
AGP, 1996 32 66 264
AGP (x2), 1996 32 132 528
NOTE
Gang of Nine The nine companies
Industry Standard Architecture (EISA) bus. that build IBM-compatible microcom-
puters which together designed the
The biggest advantage that the EISA bus had over the Micro EISA bus were AST Research, Compaq
Channel bus is that legacy ISA expansion cards could be installed Computer, Epson, Hewlett-Packard,
into an EISA bus. This feature alone was enough to sway some NEC, Olivetti, Tandy, WYSE, and Zenith
people away from Micro Channel–equipped computers, because the Data Systems. Note that only two of
Micro Channel bus did not allow ISA expansion cards to be installed. these companies, Compaq Computer
and Hewlett-Packard, are big names in
In the late 1980s, you had a choice to make when you wanted to the microcomputer industry today.
purchase a new microcomputer. You could get a microcomputer with
an EISA bus (made by anybody but IBM) and use any legacy ISA
bus expansion cards you had on hand from older systems, or you
could buy a new microcomputer from IBM with its Micro Channel
Architecture bus and buy all new expansion boards to fit into it.
The EISA bus is nearly extinct. The most common bus in Pentium-
based microcomputers is the PCI bus.
tightly tied to the 80486 that there was difficulty getting the
VL-Bus to work with the new Pentium processors. By that time,
Intel had designed its own local bus, the Peripheral Component
Interconnection (PCI) bus.
The VL-Bus is practically nonexistent in new network servers.
However, you will find two other local buses: the PCI bus and the
AGP bus.
of the PCI hot swap feature. swap. This technology allows PCI adapters to be installed, removed,
or upgraded without having to shut down the network server. This is
EXAM
a very nice feature because it means that failed PCI adapters can be
replaced without having to shut down the network server.
PCI-X
The PCI-X specification resulted from a PCI workgroup that was
led by Compaq Computer. The specification was accepted by the
PCI Special Interest Group (SIG) in 1999.
The PCI-X specification is an enhancement to the original PCI speci-
fication and it defines a 64-bit PCI bus that operates at 133MHz.
This gives a bus data transfer rate in excess of 1GB per second.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 51
Intelligent Input/Output
Intelligent Input/Output (I2O) is a technology designed to remove
the load created by input/output operations from the processor by
using specialized input/output processors (IOPs). The IOPs will
handle basic I/O operations, such as handling I/O interrupts, buffer-
ing of data between system memory and the I/O device, and the
actual data transfer. These I/O operations are currently handled by
the network server’s processor.
The I2O driver consists of two components: the OS-specific module
(OSM), which interfaces to the network operating system; and the
hardware device module (HDM), which handles communications
with the I/O device. This is done in an attempt to make the I/O
device driver independent of the network server operating system.
The first implementation of I2O will most likely be associated with
the PCI bus.
The I2O specifications, developer resources, and other materials that
were developed by the I2O Special Interest Group (I2O SIG) can be
found at www.intelligent-io.com/.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 52
InfiniBand Architecture
InfiniBand Architecture (IBA) is a proposed future high-speed
input/output system for network servers designed to replace current
bus-based input/output systems (such as PCI). IBA can use fiber-
optic cable or copper wire as a transmission medium. Transmission
speeds from 500MB per second to 6GB per second can be achieved.
IBA is being developed by the InfiniBand Trade Association, whose
membership includes Compaq, Dell, Hewlett-Packard, IBM, Intel,
Microsoft, and Sun Microsystems.
For more information about IBA, visit the InfiniBand Trade
Association web site at www.infinibandta.org.
There is a good white paper from Dell on InfiniBand Architecture at
www.dell.com/downloads/global/vectors/infiniband.pdf.
R E V I E W B R E A K
This section discussed the various system buses that a server hard-
ware specialist might expect to find in a network server. When
selecting the network server, and considering the role that the net-
work server will perform in the network, select the network server
with the fastest bus (or multiple peer buses) that fits the budget of
the network server. The system bus cannot be upgraded without
replacing the entire system board, which is an expensive proposition.
MEMORY
NOTE
There are two basic types of RAM: dynamic RAM (DRAM), and sta-
NOTE
tic RAM (SRAM). Dynamic RAM needs to be refreshed many times Future Memory Technology In the
per second to maintain data. Static RAM will maintain data stored in very near future, you should expect to
it without a refresh. Because there is no refresh cycle, static RAM is see new memory technology in net-
work servers. Mirroring of memory for
faster than dynamic RAM. Unfortunately, static RAM is rather expen-
fault tolerance and memory hot swap
sive and therefore it is not used except in special cases such as Level 2
technologies should be available very
processor cache. Dynamic RAM is utilized as main memory. shortly.
Amount of Memory
As a general rule, there is no such thing as too much memory in a
NOTE
Memory Measurements Memory is
network server. In network servers, RAM is used not only to hold measured in kilobytes, megabytes, or
programs to be executed, but also for caching frequently accessed gigabytes. However, computers use
files (in the file and print server environment), and caching fre- base 2 (or binary) arithmetic rather
than base 10 (or decimal) arithmetic.
quently accessed web pages (in the web server environment). The
Everything in computers is measured
more RAM in a network server that uses virtual memory technology
and addressed as a power of 2. Two
(also known as paging), the less likely that paging will have to occur. raised to the 10th power (210) is
Paging refers to the process of writing unused portions of program to 1024, which is close to 1000 (102).
disk to free up memory. The process of writing to a disk drive is The term “kilo” historically indicates
very slow when compared to memory access speeds. If paging in a 1000, and kilobyte is the term that is
virtual memory operating system can be avoided, the overall perfor- used to indicate 1024 bytes. In gen-
mance of the network server will be improved. eral usage outside of the computer
field, however, kilo means 1000. This
The disk storage area used by virtual memory systems is usually leads to a great deal of confusion.
called a paging file or a swap file.
Memory Interleaving
One way to speed up access to main memory is to divide it into sec-
tions so that the processor can access alternate sections at the same
time, thus avoiding having to wait for memory to finishing process-
ing one request before processing another (known as a wait state). A
two-way interleave means that main memory has been divided into
two sections. A four-way interleave means that main memory has
been divided into four sections.
Memory Types
Several types of memory are common in network servers. Memory is
categorized by several different characteristics.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 54
FIGURE 1.1
Single in-line memory module.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 55
FIGURE 1.2
A dual in-line memory module.
Memory Technologies
Memory can also be categorized by the memory technology utilized.
Important memory technologies include extended data out (EDO)
DRAM, synchronous DRAM (SDRAM), double data rate synchro-
nous DRAM (DDR SDRAM), and direct Rambus (RDRAM)
Synchronous DRAM
Instead of having its access speeds measured in nanoseconds,
SDRAM synchronizes itself with the memory bus, thus eliminating
waits between memory accesses. Common speeds are 66MHz,
100MHz, and 133MHz.
FIGURE 1.3
Direct Rambus memory module (RIMM).
06 Serv+ Ch_01 7/10/01 9:41 AM Page 56
Direct Rambus
Direct Rambus (RDRAM) is a radically different memory architecture
developed by Rambus Incorporated. It does provide very fast data
transfers compared to the more traditional SDRAM. Speeds up to
800MHz can be achieved. This high speed comes at a high cost.
However, this is the fastest memory technology currently available
for network servers.
Know the difference between data protection in that it cannot only detect single bit errors but can
Error Correcting Code and non- correct them as well.
EXAM
ECC memory.
Many desktop computer systems use neither parity nor ECC data
integrity techniques to keep the cost of the system low. In network
servers, where data integrity is critical, however, ECC memory
should be selected.
When selecting network server hardware and there is a choice
Buffered, Unbuffered, Registered between ECC memory and non-ECC memory, pick the ECC
TIP
Memory Know the meaning and memory even though it will raise the cost of the system.
purpose of buffered, unbuffered,
EXAM
NOTE
ported in the network server. Therefore, when using EDO memory, Putting It All Together Putting all
more buffered EDO memory modules can be installed in a network the memory characteristics together,
server than unbuffered EDO memory modules. you will see memory modules descrip-
tions that combine various descrip-
Buffered and unbuffered EDO memory modules cannot be mixed tors, such as the following:
in the same system. In fact the memory modules are keyed to pre- • Unbuffered SDRAM
vent the accidental insertion of the wrong type of memory module
into the memory slot. • Buffered ECC ECO DIMM
• Unbuffered EDO DIMM
For SDRAM, more registered SDRAM memory modules can be
installed in a network server than unregistered SDRAM memory • EDO SIMM
modules. Registered and unregistered SDRAM modules cannot • ECC registered DIMM
be mixed in the same system. Again, the SDRAM modules are
• ECC unbuffered SDRAM
keyed to prevent the accidental insertion of the wrong type of
memory module into the memory slot. Check the documentation that came
with the network server for the type of
For a complete description of computer memory, see the memory required for upgrading mem-
Kingston Ultimate Memory Guide (all 110 pages of it) at ory in the network server.
www.kingston.com/tools/umg/default.asp.
R E V I E W B R E A K
This section discussed the various types of system memory that the
server hardware specialist might encounter in a network server.
When selecting the memory for a network server, and considering
the role that the network server will perform in the network, pur-
chase as much memory as the budget for the network server will
allow. For network servers, it is highly recommended that ECC
memory be obtained, due to its capability to detect and correct
single bit errors in the memory itself.
Consider scalability when purchasing a new network server. How
many memory slots are available? How many memory slots will be
available after the memory specified in the original memory configu-
ration is installed? If the new network server will have 512MB of
memory installed, are there four 128MB memory modules installed
or are there two 256MB memory modules installed? Or is the
512MB of memory composed of eight 64MB memory modules?
The more memory modules installed, the fewer memory expansion
slots that will be available for the addition of memory in the future.
In fact, in the case of the eight 64MB memory modules, it might be
the case that they have to be removed and replaced entirely to
upgrade the network server to 1024MB of memory.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 58
DISK SUBSYSTEMS
. 0.7 Identify the strengths and weaknesses of the various types
of disk subsystems. Content may include the following:
• Integrated Drive Electronics (IDE/ATA)
• Enhanced Integrated Drive Electronics (EIDE/ATA2)
• Ultra ATA / Ultra DMA
• Serial ATA
• Small Computer System Interface (SCSI)
• Self-Monitoring, Analysis and Reporting Technology
(S.M.A.R.T.)
• Fibre Channel
• Fibre Channel Arbitrated Loop (FC-AL)
• Serial Storage Architecture (SSA)
• Storage area network (SAN)
• Network-attached storage (NAS)
The disk subsystem is a very important component in network
servers. Not only is the proper selection of the amount of disk stor-
NOTE
disk data transfer to memory is much more efficient than using PIO
used by IDE and EIDE technology.
The ATA adapter is still restricted to two channels with two disk
drives per channel, but using DMA disk transfers can be much
faster. There are currently three versions of Ultra ATA:
á Ultra ATA 33, or Ultra DMA 33, can transfer data at speeds
of up to 33MBps.
á Ultra ATA 66, or Ultra DMA 66, can transfer data at speeds
of up to 66MBps.
á Ultra ATA 100, or Ultra DMA 100, can transfer data at speeds
of up to 100MBps.
Ultra ATA 66 and Ultra ATA 100 require a different cable than that
NOTE
Ultra Controversy The high transfer used by Ultra ATA 33, EIDE, and IDE technologies. The cables for
speeds offered by Ultra ATA 100 are these faster modes use 80 conductors in the cable rather than 40
considered by some people to be very conductors in the cable. However, the connector used for Ultra ATA
misleading in that although it is true
66 and Ultra ATA 100 is still a 40-pin connector.
that the ATA channel can support
these very high transfer speeds, there
are currently no ATA disk drives that
can sustain such speeds. The trans-
Serial ATA
fer rate of the fastest ATA disk drives All versions of the ATA specification up to now have used a parallel
currently available is 40MBps. data transfer mechanism. (Multiple bits that represent data travel
down the ATA ribbon cable in parallel wires.) Work is under way to
create a new ATA specification that uses a serial data transfer mecha-
nism. (Bits travel down the ATA cable one after the other.) Because
serial ATA uses serial transmission, its speed is specified in megabits
per second (Mbps), whereas parallel ATA specifies its speed as
megabytes per second (MBps). The first serial ATA specification is
known as Serial ATA/1500, which means the transmission speed is
1500Mbps (1.5 gigabits per second). Because of the overhead of ser-
ial transmissions, this translates into approximately 150MBps.
The advantages of serial ATA over parallel ATA are as follows:
á A cable length of up to 1 meter (compared to 18 inches for
parallel ATA).
á A thinner (4 wire) and more flexible serial ATA cable replaces
the current flat 40-wire or 80-wire ribbon cable used by
parallel ATA.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 61
You can find additional information on serial ATA at the Serial ATA
Working Group web page at www.serialata.org.
SCSI-1
The original SCSI standard adopted by the American National
Standards Institute (ANSI) in 1979 was renamed to SCSI-1 when the
SCSI-2 standard was adopted in 1984. The original SCSI bus was 8
bits wide and could transfer up to 5MBps. The SCSI-1 bus could
have up to 7 SCSI devices attached. The maximum length of the
SCSI-1 bus was 6 meters. The SCSI-1 bus used 50-pin connectors.
SCSI-2
The SCSI-2 standard was approved by ANSI in 1994. It offered
many improvements over the SCSI-1 standard. A new physically
smaller 68-pin connector was used by SCSI-2 devices. The SCSI-2
standard defined both 8-bit-wide (which used a 50-pin connector)
06 Serv+ Ch_01 7/10/01 9:41 AM Page 62
and 16-bit-wide (which used the 68-pin connector) buses. The 16-
bit version was known as Wide SCSI, causing the 8-bit bus to be
called Narrow SCSI. A higher transfer speed, 10MBps, was specified
as Fast SCSI. Both the Fast Narrow SCSI and the Slow Wide SCSI
could transfer data at 10MBps. There was also a combination Fast
Wide SCSI that could transfer data at 20MBps. A Wide SCSI bus
can support 15 devices rather than the normal 7 devices of SCSI-1
or Narrow SCSI-2. SCSI-2 also introduced Differential SCSI (later
called High Voltage Differential (HVD) SCSI) that could have a
cable length of 25 meters.
SCSI-3
Unlike the SCSI-1 and SCSI-2 standards, the SCSI-3 specification is
NOTE
Confused? Don’t confuse Ultra 2 actually a set of several standards. SCSI-3 maintained the parallel
SCSI and SCSI 2! Ultra 2 SCSI is a bus of SCSI-1 and SCSI-2, but it also added Serial SCSI standards
SCSI-3 standard that can transfer (Fibre Channel, Serial Storage Architecture, and IEEE 1394). The
data at 40MBps. SCSI-2 is the pre-
Low-Voltage Differential (LVD) SCSI signaling system was intro-
decessor to the SCSI-3 standard that
duced with SCSI-3.
can transfer data at a maximum of
20MBps (for the Fast Wide variant). The SCSI-3 standard has led to faster SCSI buses. Table 1.2 summa-
rizes the SCSI-3 standards. Notice the marketing terms for the different
SCSI standards. SCSI-3 devices use 50-pin or 68-pin connectors. The
higher speed SCSI-3 standards use 68-pin connectors exclusively. The
internal 68-pin connector is often referred to as a high density (HD)
68-pin connector. The external 68-pin connector is often referred to as a
very high density cable interconnect (VHDCI) connector. Table 1.3 lists
SCSI cable lengths, cable types, and devices per channel.
TABLE 1.2
SUMMARY OF SCSI S TA N D A R D S
SCSI Marketing Bus Speed Bus Width Transfer
Standard Term (MHz) Speed MB/s
SCSI-1 5 8 bit 5
SCSI-2 Fast 10 8 bit 10
SCSI-2 Wide 5 16 bit 10
SCSI-2 Fast Wide 10 16 bit 20
SCSI-3 Ultra 20 8 bit 20
06 Serv+ Ch_01 7/10/01 9:41 AM Page 63
TABLE 1.3
SCSI C A B L E L E N G T H , C A B L E T Y P E , AND DEVICES
PER CHANNEL
continues
06 Serv+ Ch_01 7/10/01 9:41 AM Page 64
SCSI-3 15 68-pin - - 12
Ultra320 internal
68-pin
VHDCI
external
NOTE
the drive is about to fail, thus allowing time for the system adminis- Why Is It Spelled Fibre? The original
trator to take actions (such as performing a backup and replacing Fiber Channel specification supported
the disk drive before it fails) to avoid loss of data. fiber-optic cable only. Support for cop-
per cable was later added to the Fiber
Channel specification. It was decided
Fibre Channel to use the French spelling of Fiber
(Fibre) in the name Fibre Channel in
Fibre Channel is an ANSI standard serial data transfer architecture. hopes that this would reduce the nat-
In one form, it is a version of serial SCSI-3. In another form, it can ural connection of Fibre Channel with
use TCP/IP to transfer data. Fibre Channel is an external bus that fiber-optic cable. A Fibre Channel can
allows tens, hundreds, and thousands of devices, such as disk drives, be implemented using either fiber-
to be connected to a single Fibre Channel. Fibre Channel can trans- optic cable or copper cable.
mit data at up to 2GBps using fiber-optic cable.
Fibre Channel can be implemented using one of several different
topologies: the fabric topology, the point-to-point topology, and the
arbitrated loop topology. The most common of these topologies is
the arbitrated loop topology.
For more information about Fibre Channel, see the Fibre Channel
Industry Association (FCIA) web site at www.fibrechannel.com.
Network-Attached Storage
As the name implies, network-attached storage (NAS) is disk or tape
storage attached to a network that is accessed via the network itself
using the common network systems (for instance, Ethernet or Token
Ring) and common network protocols, such as the Network File
System (NFS), the Common Internet File System (CIFS) for
Windows networks, FTP, and HTTP. Figure 1.5 shows an example
of a NAS system. Network clients communicate directly with the
intelligent storage server.
Local Area Network using standard network protocols (TCP/IP, IPX/SPX, NetBEUI)
Disk
FIGURE 1.4 Disk
Disk Disk Disk Disk
Storage area network. Disk Disk Disk
06 Serv+ Ch_01 7/10/01 9:41 AM Page 67
R E V I E W B R E A K
This section discussed the disk subsystem for a network server.
When selecting the disk subsystem for a network server and consid-
ering the role that the network server will perform in the network,
acquire the fastest disk drives available. When planning for storage
requirements, remember that there is no such thing as too much
disk storage on a network server. Plan for expandability. Consider
the number of disk drives that will fit into the network server
chassis, as well as the number of disk drives that can be attached
to a single controller bus.
In general, SCSI disk drives offer the most expandability and the
fastest transfer rates. However, Ultra ATA disk drives are good choices
for entry-level servers that can tolerate the limited number of disk
drives (four) that can be attached to a single ATA disk controller.
Always buy disk drives that support S.M.A.R.T. technology.
S.M.A.R.T technology can be found on both ATA and SCSI
disk drives.
continues
06 Serv+ Ch_01 7/10/01 9:41 AM Page 68
continued
Because disk drives are the only mechanical devices (other than cool-
ing fans) in a network server, there is a good chance that some of
them will fail during the useful life of the network server. Plan for
fault tolerance of the disk subsystem. Always use RAID technology
on network servers; hardware RAID has many advantages, but does
cost more to implement than software-based RAID. Software-based
RAID does use some of the processing power of the network server
processor, but any RAID is better than no RAID at all.
NETWORK SYSTEMS
NOTE
NOTE
of 10Mbps and uses a media access method known as CSMA/CD DIX Connector The 15-pin connector
(carrier sense multiple access with collision detection). Ethernet is that was standard on most Ethernet
the most popular networking system in the world. adapters until a few years ago is
often referred to as a DIX connector
Faster versions of Ethernet have been created since the original in honor of the three companies that
Ethernet standard was finalized. Fast Ethernet (802.3u), which oper- proposed the Ethernet standard
ates at 100Mbps, and Gigabit Ethernet (802.3z), which operates at (Digital, Intel, and Xerox). The correct
1000Mbps, are the names of the new, faster Ethernet standards. name for this connector is the
Attachment Unit Interface (AUI) con-
nector. No wonder it was called the
Standard Ethernet DIX connector.
The following sections cover the 10Mbps Ethernet standards. The
original Ethernet is now known as 10BASE-5. The next version of
Ethernet that used a much thinner cable is known as 10BASE-2. For
many years these were the only Ethernet standards. However, the need
NOTE
Project 802? The IEEE project 802
for cheaper cabling prompted the creation of the 10BASE-T Ethernet was so named because it was started
standard using twisted-pair cabling. The use of fiber-optic cable for in February of 1980.
increased cable length led to the 10BASE-F Ethernet standard.
10BASE-5
The original Ethernet used a thick coaxial cable (RG-8) to connect
computers. The maximum cable length was 500 meters. The name
10BASE-5 indicates that the speed is 10Mbps, it uses baseband
transmission, and the maximum cable length is 500 meters. A
10BASE-5 network is wired as a bus (a long length of cable) with
computers attached at various points along the length of the cable.
10BASE-2
The thick coaxial cable used by the original Ethernet was not only
costly, but also very difficult to work with due to its thickness. The
10BASE-2 specification for Ethernet uses a thinner coaxial cable
(RG-58) that is more flexible and cheaper—hence the names thinnet
and cheapernet, which are often used to refer to 10BASE-2. As the
name implies, 10BASE-2 has a speed of 10Mbps, uses baseband
transmission, and has a maximum cable length of 185 meters. Yes, it
is 185 meters, not 200 meters as the name implies. Like 10BASE-5,
10BASE-2 is also wired as a bus.
10BASE-T
The 10BASE-T specification for Ethernet changed the Ethernet
world forever. Rather than coaxial cable, 10BASE-T uses
unshielded twisted-pair (UTP) cable that is similar to phone cable.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 70
10BASE-F
The 10BASE-F specification is for 10Mbps Ethernet running over
fiber-optic cable rather than copper wire. Maximum cable length is
2 kilometers.
IN THE FIELD
BIT SIZE
Plugging these numbers into the formula, you get the following bit
lengths for 10Mbps Ethernet.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 71
Fast Ethernet
The Fast Ethernet standard changes the speed of Ethernet from
10Mbps to 100Mbps. Fast Ethernet can use twisted-pair cable or
fiber-optic cable. The IEEE specification for Fast Ethernet is 802.3u.
Fast Ethernet can use Category 5 twisted-pair cables up to 100
meters long.
IN THE FIELD
continues
06 Serv+ Ch_01 7/10/01 9:41 AM Page 72
continued
• Category 3 (Cat 3), 16MHz, 4.8–9.6 twists/foot
Used for: 10BASE-T, 4Mbps Token Ring
• Category 4 (Cat 4), 20MHz, 12–20 twists/foot
Used for: 10BASE-T, 16Mbps Token Ring
• Category 5 (Cat 5), 100MHz, 15–24 twists/foot
Used for: 10BASE-T, 100BASE-T, 1000BASE-T, 16Mbps
Token Ring
• Category 5E (Cat 5E), 100MHz, 15–24 twists/foot
Used for: 10BASE-T, 100BASE-T, 1000BASE-T
• Category 6 (Cat 6), 250MHz (proposal)
100BASE-T
The 100BASE-T specification is actually three separate specifica-
tions: 100BASE-TX, 100BASE-T4, and 100BASE-F.
100BASE-TX uses the same wiring scheme as 10BASE-T, but the
unshielded twisted-pair cable must meet the Category 5 specification
for it to work.
100BASE-T4 uses a twisted-pair cable with 4 pairs (8 wires), but the
twisted-pair cable can be Category 3, Category 4, or Category 5.
Very few companies make equipment, hubs and NICs, that can use
the 100BASE-T4 standard.
100BASE-F
100BASE-F is the specification for Fast Ethernet over fiber-optic
cable and it allows data transmission at 100Mbps for a maximum
cable length of up to 2 kilometers.
Gigabit Ethernet
There are specifications for Ethernet that are faster than Fast
Ethernet. Instead of calling the next Ethernet standard Faster
Ethernet, the name Gigabit Ethernet is used. Gigabit Ethernet
can transfer data at 1Gbps (or 1000Mbps). The Gigabit Ethernet
standard was ratified in 1998.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 73
1000BASE-T
Gigabit Ethernet using twisted-pair wiring is known as 802.3ab.
Gigabit Ethernet uses all four pairs of wires in a high-quality
unshielded twisted-pair (Category 5 or higher) cable. Maximum
cable length is 100 meters.
1000BASE-SX
The 1000BASE-SX specification is a 1-gigabit Ethernet standard
using fiber-optic cable. It is designed for low-cost fiber-optic cable
over distances of 220 meters to 550 meters. It comprises part of the
802.3z specification.
1000BASE-LX
The 1000BASE-LX specification is a 1-gigabit Ethernet standard
using fiber-optic cable that is designed for long distances using more
expensive single-mode fiber-optic cable over distances of up to 5
kilometers. It is also part of the 802.3z specification.
10 Gig Ethernet
The proposed 10-gigabit per second Ethernet standard is known as
802.3ae. For more information about the 10 Gig Ethernet standard,
check out the web site at www.10gea.org.
10 Gig Ethernet will use only fiber-optic cable as a transmission
medium. It is anticipated that the new standard will allow data
transmission speeds of 10Gbps for distances of 100 meters using
multimode fiber-optic cables and distances of at least 2 kilometers
using single-mode fiber-optic cable.
Adapter Teaming
Multiple NICs are often found in network servers. In the past, how-
ever, multiple NICs were attached to different physical networks,
making the server available on all the attached networks. Although
06 Serv+ Ch_01 7/10/01 9:41 AM Page 75
multiple NICs might still serve the same purpose today, a technique
known as adapter teaming provides new capabilities.
Adapter teaming is the technique of using more than one NIC in a
network server to provide adaptive fault tolerance, adapter load bal- Multiple NICs Know the
TIP
ancing, or port aggregation. The network server operating system terms that deal with multiple
NICs in a network server: adapter
must support adapter teaming for it to be implemented on the net-
EXAM
teaming, adaptive fault tolerance,
work server. The following list describes the various capabilities that
adapter load balancing, and port
can be provided by using various forms of adapter teaming: aggregation.
á Adaptive or adapter fault tolerance. Allows multiple NICs to
be installed in a network server, all of which are attached to
the same network. Only one NIC, the primary NIC, will be
active. A secondary NIC in the team will be utilized when
there is a failure of the primary NIC in the team. The NICs in
the team may be running at different speeds.
á Adapter or adaptive load balancing. Allows multiple NICs to
be installed in the network server (up to eight), all of which are
attached to the network. The primary NIC will both receive
and transmit data, whereas the other NICs in the team share
the transmission load. All NICs in the team should run at the
same speed. If the primary NIC fails, one of the other NICs
will take over its function, thus providing fault tolerance.
á Port aggregation. Allows multiple NICs to act together to
NOTE
increase the bandwidth available to the network server. Magic Packet A magic packet is a
Requires an Ethernet switch that supports port aggregation. standard TCP/IP packet that contains
For example, four 100Mbps NICs could be used to provide the IP address of the destination com-
puter with WOL capabilities repeated
400Mbps total bandwidth to the network server.
16 times in the data portion of the
packet. Upon receipt of the magic
packet, the WOL NIC powers up the
Wake-On-LAN computer in which it is installed.
A recent addition to the functions that can be performed by a NIC
is called Wake-On-LAN (WOL). Wake-On-LAN technology enables
a network administrator to power up a computer with a WOL NIC
from anywhere on the network. A special TCP/IP packet, called a WOL and Its Magic Packet Know
TIP
magic packet, is sent to the NIC on the remote computer, which will how Wake-On-LAN and its magic
then power up the computer. This feature is handy for remote trou- packet operate.
EXAM
Network Devices
Several different network devices can be used to create a network. As
a server hardware specialist, you should be familiar with the function
of these devices.
Repeater
A repeater is a network device that allows the use of longer cables
than specified by the network standards. In a 10BASE-5 Ethernet
network, which normally has a maximum single cable length of 500
meters, for instance, two 500 meters cables can be joined together
by a repeater to form a total network length of 1000 meters. The
repeater receives the signals from one cable segment and regenerates
the signals and sends them to the other cable segment. A repeater is
not an amplifier. Amplifying the signals would not only amplify the
data signal but the noise as well. That is why a repeater regenerates
the signal. A repeater can also act as a media converter. A coaxial
cable could be connected to one port and a fiber-optic cable could
be connected to the other port.
Hub
A hub is a network device used in an Ethernet network that uses
twisted-pair cabling. Each network computer is attached to the hub.
The hub is nothing more than a multiport repeater.
Bridge
A bridge is a network device that segregates two segments on the same
network. Bridges are often used to connect a network in a building to
a backbone network. The bridge will keep network traffic that is des-
tined for computers within the building local and only forward net-
work traffic to the backbone when the destination is a computer not
in the building. A bridge makes its decision based on the hardware
addresses of the source and destination computers. A bridge can be a
hardware device or a network server in the role of a bridge.
Switch
A switch is a network device that is basically a very high-speed
bridge. It creates a temporary link between the source and destina-
tion computers (based on hardware addresses) for the time it takes to
transmit a single data packet. Multiples of these temporary links can
06 Serv+ Ch_01 7/10/01 9:41 AM Page 77
Router
Network Devices Know the func-
TIP
A router is a network device that moves data packets from one net- tion of the various network devices:
work to another based on the network address of the destination repeaters, hubs, bridges, switches,
EXAM
computer. The network address is also sometimes called the software and routers.
address. TCP/IP addresses and IPX/SPX addresses are software
addresses. A router can be a hardware device or a network server
operating in the role of a router.
Network Protocols
Network protocols are sets of rules that allow data and messages to go
from one computer to another over the network. There are literally
hundreds of network protocols, but luckily only a few of these are in
common use. The three most common network protocols are
TCP/IP, IPX/SPX, and NetBEUI. Other protocols that might be
encountered on a network include AppleTalk and DLC.
TCP/IP
TCP/IP (Transmission Control Protocol/Internet Protocol) is cur-
rently the 500-pound gorilla of network protocols. TCP/IP is avail-
able on practically all computer systems, from the smallest home
computer to the largest super computer. Even successful network
operating systems, such as Novell NetWare, that had used their own
network protocols in the past, have switched to TCP/IP. However,
more important than the popularity of TCP/IP is the fact that it is
the protocol of the Internet. If you want to surf the web, you’ve got
to have TCP/IP.
IPX/SPX
The Internet Packet eXchange/Sequenced Packet eXchange (IPX/SPX)
protocol is a derivative of Xerox’s XNS (Xerox Network System)
protocol. It was the only protocol used with early versions of Novell
NetWare (versions 2 and 3 and 4) and is an option on NetWare
06 Serv+ Ch_01 7/10/01 9:41 AM Page 78
NetBEUI
NetBIOS Enhanced User Interface (NetBEUI) was developed
by IBM. The NetBIOS acronym stands for Network Basic
Input/Output System. So the full name would then be Network
Basic Input/Output System Enhance User Interface. Don’t you
just love it when acronyms are used to make other acronyms. IBM
used NetBEUI for early PC networks in its LAN Server product,
believing that PC networks would generally have fewer than 20
microcomputers on them.
The NetBIOS naming component of NetBEUI is still utilized in
Microsoft operating systems, even if the underlying protocol is
TCP/IP or NWLink. Microsoft developed NetBIOS over TCP/IP
and NetBIOS over NWLink to allow its network naming system to
operate regardless of the protocol being used on the network. Even
Novell had a version of NetBIOS over IPX in its NetWare operating
system to support the applications that used the early NetBIOS
naming convention.
Microsoft used NetBEUI for LAN Manager, Windows for
Workgroups, Windows 95, Windows 98, and Windows NT
operating systems. However, all of these Microsoft operating
systems supported other protocols as well. For backward
compatibility, it is available in Microsoft Windows 2000 as well.
NetBEUI works well only for small networks consisting of fewer
than approximately 20 computers. It is reliable and self-tuning, but
it cannot be routed. The lack of routability means that large net-
works cannot be built using NetBEUI as the protocol.
AppleTalk
The AppleTalk protocol was developed by Apple Computers to con-
nect Macintosh computers together in a small network. The original
AppleTalk could support up to 32 devices, over a distance of 1000
feet. The communications speed is 230,000 Kbps.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 79
DLC
The Data Link Control (DLC) protocol cannot be used to allow
computers on a network to communicate with each other. Rather it
is used as a mechanism to allow IBM mainframes to communicate
with microcomputers on a network.
The DLC protocol can also be used to communicate with network-
attached printers.
Network Management
As networks get larger and more complicated and the number of
qualified people available to manage the networks remains limited,
the need for software tools to help manage the network increases
dramatically. Luckily, several software network management tools
are available to make network management possible. The oldest and
most popular of these tools is the Simple Network Management
Protocol (SNMP), which is part of the TCP/IP protocol suite.
Nearly all network devices, including network servers and their
software, can be managed via SNMP, provided that SNMP agent
software is available and it has been installed. Several other manage-
ment technologies are also available, including the Desktop
Management Interface (DMI), the SCSI Accessed Fault-Tolerant
Enclosures (SAF-TE) technology, and SCSI Enclosure Services
(SES). All of these technologies are described on the following pages.
TIP
displayed locally on the enclosure itself by means of status lights.
SAF-TE uses the SCSI channel itself as the communication mecha- and specifications of the SCSI
Accessed Fault-Tolerant Enclosures.
nism. SAF-TE can also generate SNMP alerts that can be sent to an
EXAM
SNMP management station on the network.
You can find a white paper on the SAF-TE specification at
www.andataco.com/support/whitepapers/safte.pdf .
R E V I E W B R E A K
This section discussed the network subsystem of the network server.
Which network technology (Ethernet or Token Ring) was one of the
topics of the holy wars of computing and networking in the early
1990s. It appears that Ethernet won that battle, as all new develop-
ment for faster networks has been using Ethernet.
With the advent of the use of twisted-pair wiring to implement net-
works, they are now much simpler to implement than they were in
the days of coaxial cable.
The protocol of choice today is TCP/IP, due to its flexibility and the
fact that it is the protocol used to access the Internet. TCP/IP is
available for most computers from the smallest desktop computer to
the largest supercomputer. However, TCP/IP does require more con-
figuration that most of the other protocols; but the use of DHCP
and APIPA has made that much less painful.
The use of SNMP to manage not only network devices, such as
routers and hubs, but also network servers, software applications
(such as database or email), and desktop computers has made the
task of managing large installations much easier. However, devices
that support SNMP agents still cost more than devices that do not
support SNMP agents. The SNMP management software is also
rather expensive to acquire.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 82
BACKUP HARDWARE
. 0.9 Identify various types of hardware used for backup systems.
Backup hardware types include:
• Tape drives
• Disk drives
• Other backup devices
To perform a backup (copying of the data and software on a network
server), some backup hardware must be available on the network
server. The most common backup hardware device (historically and
now) is some form of magnetic tape drive. Other backup hardware
devices can be used for backup operations, but these are much less
common than tape drives.
A backup hardware device is a required part of any network server
installation. Backup hardware devices can be expensive to purchase.
However, the consequences of not having a backup hardware device
on a network server, and thus not having a backup copy of the data
and software of the network server, can be far more expensive.
Tape Drives
Tape drives are most commonly used as the device for backup of the
data on a network server’s disk drives. There are a variety of tape
devices that use different tape formats for storing data. Many tape
drives can also compress the data before it is stored on the tape.
In most cases the compression ratio is 2:1. This has the effect of
doubling the storage capacity of the tape.
The QIC standard has limited storage capacity and is used only in
entry-level network servers.
TABLE 1.4
QIC T A P E S TA N D A R D S
QIC QIC Cartridge Storage Capacity Interface
Standard Native/Compressed
QIC-40 DC-2000 40MB/80MB Floppy
QIC-40 DC-2060 60MB/120MB Floppy
QIC-80 MC-2120 125MB/250MB Floppy, parallel
QIC-80 MC-2120Extra 400MB/800MB Floppy, parallel
QIC-80XL MC-2120XL 170MB/340MB Floppy, parallel
QIC-3020XL MC-3020XL 680MB/1.36GB Floppy,
parallel, IDE
QIC-3020XL MC-3020Extra 1.6GB/3.2GB Floppy,
parallel, IDE
QIC-3095 MC-3095 4GB/8GB IDE, SCSI-2
QIC-3220 MC-3220 10GB/20GB SCSI-2
QIC-5010 DC-5010 16GB/32GB SCSI-2
TABLE 1.5
T R AVA N T A P E D R I V E S TA N D A R D S
Travan Tape Cartridge Storage Capacity Interface
Native/Compress
Travan-1 TR-1 400MB/800MB Floppy, parallel
Travan-2 TR-2 800MB/1.6GB Floppy, parallel
Travan-3 TR-3 1.6GB/3.2GB Floppy, parallel
Travan-4 TR-4 4GB/8GB SCSI-2, EIDE
Travan NS-8 NS-8 4GB/8GB SCSI-2, EIDE
Travan-5 TR-5 10GB/20GB SCSI-2, EIDE
TABLE 1.6
T R AVA N R E A D /W R I T E AND R E A D C O M PAT I B I L I T Y
Travan Standard Can Read/Write Can Read
Travan-1 QIC-80 QIC-40
Travan-3 TR-2, QIC-3010, QIC-80, TR-1
QIC-3020
Travan-4 QIC-3095 QIC-3010, QIC-3020,
QIC-80, TR-3, TR-1
Travan NS-8 QIC-3010, QIC-3020,
QIC-80
8mm Tape
Tape technology that uses 8mm tape was pioneered by Exabyte
Corporation. The technology uses a tape similar to 8mm videotape
and the same helical scan system used by a VCR. Table 1.7 reviews
8mm tape technologies.
TABLE 1.7
8 M M TA P E T E C H N O L O G I E S
8mm Tape Technology Storage Capacity Transfer Speed
Native/Compressed
8200 2.5GB/5GB 246KB per second
TABLE 1.8
M A M M O T H 8 M M TA P E T E C H N O L O G I E S
Mammoth Technology Storage Capacity Transfer Speed
Native/Compressed
Mammoth-1 20GB/40GB 3MBps
TABLE 1.9
AIT T A P E S TA N D A R D S
AIT Standard Tape Media Storage Capacity Transfer Speed
Native/Compressed
AIT-1 SDX125C 25GB/50GB 3MBps
AIT-1 SDX135C 35GB/70GB 3MBps
AIT-2 SDX236C 36GB/72GB 6MBps
AIT-2 SDX250C 50GB/100GB 6MBps
For more information about AIT technology, see the AIT Forum
web site at www.aittape.com.
TABLE 1.10
DAT T A P E S TA N D A R D S
DDS Format Storage Capacity Transfer Speed AKA
Native/Compressed
DDS-1 2/4GB 1MBps --
DDS-2 4/8GB 1MBps DAT8
DDS-3 12/24GB 2MBps DAT24
TIP
Digital Linear Tape (DLT) technology offers high-capacity and rela- storage capacities of the different
tively high-speed tape backup capabilities. DLT tapes record infor- DLT formats.
EXAM
mation on the tape in a linear format unlike 8mm tape technologies
that use helical scan recording techniques. DLT tape drives support
high storage capacity and a fast transfer speed. However, DLT tape
drives are expensive. Table 1.11 compares DLT tape formats.
TABLE 1.11
DLT T A P E S TA N D A R D S
DLT Standard Storage Capacity Transfer Speed
Native/ Compressed
DLT-2000 10GB/20GB 1.25MBps
DLT-2000XT 15GB/30GB 1.25MBps
DLT-4000 20GB/40GB 1.5MBps
DLT-7000 35GB/70GB 5MBps
DLT-8000 40GB/80GB 6MBps
Linear Tape-Open
Linear Tape-Open (LTO) technology was developed by Hewlett-
Packard, IBM, and Seagate. LTO comes in two distinct forms: one
designed for high storage capacity (Ultrium), and one designed for
fast access (Accelis). Table 1.12 reviews the LTO tape formats.
TABLE 1.12
LTO T A P E F O R M AT S
LTO Format Storage Capacity Transfer Speed
Native/Compressed Native/Compressed
Ultrium 100GB/200GB 20MBps/40MBps
For more information about LTO tape technology, see the LTO
web site at www.lto-technology.com.
Tape Arrays
Several network server vendors offer an array of tape drives with
fault-tolerance characteristics. Most of these technologies use four
identical tape drives and implement the tape version of RAID,
which is called RAIT (redundant array of independent tapes). RAIT
can be used for mirroring of tape drives, or (with at least three tape
drives) implemented as data striping with parity. The result is that if
a tape is damaged or lost, data recovery can still occur.
Tape Autochangers
A tape autochanger (also known as a tape auto loader) allows the tape
drive to load a new tape when the current tape gets full while perform-
ing a backup. This relieves the operator from having to remove one
tape and insert a new tape. This is very handy because backups are usu-
ally performed in the dead of night. Most tape autochangers support
unloading and loading a limited number of tapes (10 or fewer).
Tape Libraries
A tape library is usually an external system that has multiple tape
drives, tens or hundreds of tapes, and an automatic mechanism for
locating the tapes, loading them into the tape drives, and returning
the tapes to the proper location. With all this intelligence and
sophistication, it should be obvious that tape libraries are the high
end of backup systems. This can be translated into meaning that
they are expensive.
Disk Drives
Disk drives can also be used for backup operations. Performing a
backup to disk drives is a faster operation than performing a backup
to a tape drive. However, unless the “backup disk drives” can be hot
swapped, there is no provision for off-site storage of the backup.
Having the backup hard disk drive in the same room as the network
server doesn’t provide much protection in the event that the network
server room is destroyed by fire.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 89
R E V I E W B R E A K
This section discussed the various types of backup hardware available
for a network server. The most common device used for backup is a
magnetic tape drive. There is a huge assortment of magnetic tape
drives from which to choose. It is best to choose a tape drive that has
enough capacity to back up the entire network server on a single tape.
However, this is not always possible. If multiple tapes are going to be
required to backup the network server, consider obtaining a tape drive
with an autochanger (unless you really want to be in the server room
at 3 a.m. to change out the tape while the backup is running).
Although tape capacity is important, so is the speed at which data is
written to the tape. A high-capacity tape that takes 8 hours to back
up the network server might not be a wise choice, if you have only
4 hours in which to perform the backup. A lower capacity tape with
an autochanger, which can perform the backup in 2 hours, would be
a much better choice.
For increased capacity and speed, some tape drives can be purchased
in arrays that perform much like disk arrays. Data is written to sev-
eral tapes simultaneously. Some tape arrays even provide fault toler-
ance, which means that the data can be recovered even if a tape in
the array has been damaged, lost, or destroyed.
REDUNDANT COMPONENTS
. 0.10 Identify hardware components that might fail and plan
on redundancy of those components. Content may
include the following:
• Power supplies
• Fans
• Hot swap
06 Serv+ Ch_01 7/10/01 9:41 AM Page 90
• Hot plug
• Fail-over
• Hot spare
• NIC fail-over
To keep network servers operational around the clock, you can
use redundant components in the network server in several
different schemes.
Redundant Fans
Network servers can generate a large amount of heat internally, due
to the multiple processors, large amount of memory, and number of
internal disk drives. This heat must be removed from the network
server chassis or heat-sensitive network server components can fail.
This heat removal is accomplished by means of cooling fans. If a
cooling fan fails, however, the network server will build up heat
internally and either fail or be shut down by the heat-monitoring
features built in to many network servers. Some network servers
have multiple cooling fans placed front to back to provide some
redundancy in case of failure of the cooling fan. Many cooling fan
systems allow for the failed cooling fan to be replaced while the net-
work server is operational. This capability is called hot pluggable fans.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 91
Hot Swap
Hot swapping is the capability to replace components in a network
server while the server is running. Hot swapping power supplies
and hot swapping cooling fans (as mentioned above) serve as good
examples of hot swapping. Disk drives can also be acquired that
allow hot swapping. The terms “hot swap” and “hot plug” are often
confused and misused.
Hot Plug
Hot plugging is the capability to add, replace, or remove components
while the system is operational. Some PCI buses support hot plug-
ging of PCI adapters. The Universal Serial Bus (USB) supports hot
plugging of USB devices while the system is running. Hot plugging
is more sophisticated than hot swapping in that a device that is hot
plugged is automatically recognized by the system.
Fail-Over
Fail-over is a technology that allows for multiple components to be
installed in a network server, but only one of the components is
operational at any given time. When the one component fails, the
network server automatically switches over to the “fail-over” compo-
nent and keeps running.
Hot Spare
A hot spare is a network component that is powered up but not cur-
rently being used (usually a disk drive). Hot spares are often found
in RAID systems. If a disk drive fails in the RAID array, the RAID
controller starts rebuilding the data from the failed disk drive on the
hot spare drive.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 92
NIC Fail-Over
The network server can have two NICs installed, but the NIC driver
uses only one. If the primary NIC fails, the NIC driver automati-
cally starts using the other NIC (fail-over to the other NIC).
R E V I E W B R E A K
This section discussed the various redundant components that the
server hardware specialist might find in a network server. The use of
redundant components in a network server helps provide fault toler-
ance. Without redundancy, even failure of simple components, such
as cooling fans, can mean downtime of the network server. High
availability requires the use of redundant components. Of course the
high availability provided by redundant components does mean that
the price of the network server will increase.
SERVER CLUSTERING
. 0.11 Define server clustering. Describe the hardware compo-
nents required to cluster servers.
Server clustering is having multiple independent network servers per-
form as a single system. This configuration can provide high availabil-
ity and scalability. If one of the network servers in a server cluster
fails, another network server in the cluster will take over its workload.
To create a server cluster, several things are needed, including at least
two network servers, some type of hardware that will allow the net-
work servers to share disk drives, and some special software designed
to implement the cluster. Server clustering is not a new technology.
Digital Equipment Corporation (DEC) created clusters of its VAX
computers in the 1970s.
The critical part of creating a cluster of servers is the method used to
allow access to the disk drives among the network servers. There are
currently three basic methods:
á Mirrored disk drives. The characteristics of this method are
that each network server has its own disk drives, but the data on
the disk drives are copied to at least one other network server in
the cluster. The problem with disk mirroring is keeping the
copies of the data on the mirrored disk drives synchronized.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 93
SCALABILITY
. 0.12 Identify hardware components that give a network server
the capability to scale up in order to meet the demands
of a larger workload.
Scalability is the capability to “grow” a network server. Scalability
is planned at the time of the purchase of the network server. If
it is suspected or expected that the load on the network server will
increase with time, it is best to buy a network server that is scalable.
Examples of scalability include the following:
á The capability to easily install additional processors in the
network server
á The capability to easily add much more memory to the
network server
á The capability to easily add more disk drives to the network
server chassis
06 Serv+ Ch_01 7/10/01 9:41 AM Page 94
HIGH AVAILABILITY
. 0.13 Identify hardware components that give a network server
the capability to be up a very high percentage of the
time, up to and including a 24 hour per day, 7 day per
week operating environment.
High availability is the goal of keeping the network server opera-
tional 24 hours per day, 7 days per week. Many techniques can be
used to try to achieve high availability. In addition to methods
already discussed (redundancy, fault tolerance, RAID, UPS, cluster-
ing), some other methods are also available.
Some network server vendors provide what they call high availability
solutions. Most of these involve two servers sharing an external set of
RAID disk drives. When one server fails, the other server takes over.
This high availability solution is expensive because one server is not
utilized at all until the other server fails.
Some clustering schemes have the capability to keep running if one
network server in the cluster fails. This is a much better solution in
that all servers are operational and providing services all the time.
The loss of one network server in the cluster will degrade perfor-
mance, but the system supported by the cluster keeps running.
BACKUP STRATEGIES
. 0.14 Identify various strategies to perform backup of the data
on the network server and plan for the recovery of the
data in the event of a hardware failure.
• Full backup
• Incremental backup
• Differential backup
A copy of the data stored on the network server is a very valuable
asset when there is a network server failure or you must recover a file
that was “accidentally” deleted by a user. A backup strategy (making
copies of the data on the network server) is a critical part of imple-
menting a network server.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 95
NOTE
Backup Sins The four big sins asso-
activity on the network server). During the week, either incremental ciated with backups are as follows:
backups or differential backups are usually performed. Whether the • Not having a backup system or
backup scheme is differential or incremental depends on how much plan
time is available to perform the back up and how quickly the data
• Having a backup system and
needs to be restored in the event of a loss of data. Incremental back-
plan, but not performing regular
ups take less time each day to backup the data, but take longer to backups
restore after a loss of data. Differential backups take longer to per-
form the backup, but allow for a quicker restore operation after a • Having a backup system and plan,
loss of data. performing backups on a regular
schedule, but not ever doing a
For more information on backup strategies, see the section trial restore to make sure that the
“Understanding Backups” in Chapter 5. data on the backups can be read
correctly
and 5 use “parity” information that is calculated from the bit patterns
of the data being written to the RAID array to provide fault tolerance.
When a disk drive fails in RAID 3, 4, or 5, the parity information can
be used along with the data on the remaining disk drives in the array
to calculate the data that was on the disk drive that failed. This allows
the disk subsystem and the network server to keep functioning, albeit
a bit slower due to the calculations required to “re-create” the missing
data. RAID level 2 is the oddball RAID in that it doesn’t use duplica-
tion or parity to provide fault tolerance. RAID 2 uses a special “ham-
ming code” instead.
RAID is a term that is surrounded by a tremendous amount of mis-
information. There is disagreement about how many levels of RAID
are defined, whether the “I” in RAID stands for inexpensive or inde-
pendent, and whether the “A” in RAID stands for array or arrays.
You’ll find RAID 6, 7, 10, 50, 53, and others mentioned in many
vendors’ literature. This section attempts to end some, if not all, of
this misinformation.
RAID was defined in 1987 in the paper “A Case for Redundant
Arrays of Inexpensive Disks (RAID)” written by David A. Patterson,
Garth A. Gibson, and Randy H. Katz at the University of
California, Berkeley. The original paper defined five levels of RAID
(1, 2, 3, 4, and 5) and offered the RAID solution as an alternative to
SLED (single large expensive disk). The original paper can be
viewed at http://sunsite.berkeley.edu/Dienst/UI/2.0/
Describe/ncstrl.ucb/CSD-87-391.
In the past few years, many people have started using the phrase
redundant array of independent disks, substituting the word “inde-
pendent” for “inexpensive.” This is very ironic in that, in 1987 disk
drives were anything but inexpensive.
RAID 0
RAID 0 was not defined in the 1987 Berkeley paper. In fact, it
NOTE
RAID level 0.
Disk Controller
RAID Level 0 -
Stripe Set WITHOUT Parity
RAID 1
RAID 1 requires a minimum of two disk drives (all other RAID lev-
els, except level 0, require at least three disk drives) to implement.
RAID 1 writes all data to two separate locations. To store 20GB of
data using RAID 1, two 20GB disk drives are required. This is a
50% loss of storage capacity.
There are two ways to implement RAID 1: through disk mirroring
or disk duplexing.
In disk mirroring, the two disk drives are connected to the same disk
controllers. The only problem with disk mirroring is that if the disk
controller fails, there is no access to the mirrored data. Figure 1.7
shows a diagram of disk mirroring. To eliminate this single point of
failure, use disk duplexing rather than disk mirroring.
Drive 1 Drive 2
Disk Controller
In disk duplexing, each disk drive in the mirrored set is connected to Drive 1 Drive 2
to the storage capacity of one of the disk drives. If you have three
10GB disk drives in a RAID 5 array, for instance, the storage capac-
ity of the array will be 20GB (a loss of 1/3, or 33%).
RAID 5
RAID 5 uses block-level parity, but spreads the parity information
among all the disk drives in the disk array. This eliminates the parity
drive failure common in RAID 4 systems. The loss of storage capac-
ity in RAID 5 systems is equivalent to the storage capacity of one of
the disk drives. If you have three 10GB disk drives in a RAID 5
array, the storage capacity of the array will be 20GB (a loss of 1/3, or
33%). In another example, if you have seven 10GB disk drives in a
RAID 5 array, the total storage capacity of the array will be 60GB
(a loss of 1/6, or 16.67%). Figure 1.9 shows a diagram of RAID 5.
RAID 0/1
RAID 0/1 is also known as RAID 0+1 and it is sometimes called RAID
10. RAID 0/1 requires at least four disk drives to implement. In RAID
0/1, there are two RAID 0 stripe sets (used to provide high input/out-
put performance) that are mirrored (which provides the fault toler-
ance). Figure 1.10 shows a diagram of RAID 0/1.
RAID Controller
RAID controllers are specialized disk controllers that use either ATA or
SCSI technologies. ATA RAID controllers are limited in the number of
disks that can be attached due to ATA channel limitations, which are a
maximum of two channels with a maximum of two disk drives per
Block 6 Parity 2
Disk Controller
RAID 0/1
Mirrored Stripe Sets Disk 4 Disk 5 Disk 6
channel (for a total of four disk drives). SCSI RAID controllers have
multiple channels. (Two channels are common, RAID controllers with
three, four, and five channels are available.) RAID controllers are gener-
ally expensive due to the sophistication that they must contain.
RAID controllers often have an onboard memory cache ranging in
size from 4MB to 256MB. This onboard memory cache often has a
battery backup system to prevent data loss in the event of sudden
power loss to the network server. This is critical because data written
from the system memory to the RAID controller is first written to
the onboard cache and it could be several seconds before the data is
actually written to disk. Without the battery to supply power to the
RAID controller, onboard memory cache could lead to the data on
the disk drive not being updated with the current data. This could
easily lead to the loss of data integrity.
The memory cache on the RAID controller can usually be config-
ured as read cache, write cache, or a combination of both. The read
cache will improve the read performance. The write cache will allow
the processor to continue with other tasks instead of waiting for the
data to be written to the disk.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 102
R E V I E W B R E A K
This section discussed the various methods available to give fault-
tolerant capability to a network server. Network servers should uti-
lize RAID technology. Hardware-based RAID is better than
software-based RAID, but software-based RAID is much better than
no RAID at all.
DISASTER RECOVERY
. 0.16 Identify techniques and plan for recovery from a disaster
(total failure of the network server).
A written disaster-recovery plan should be a standard part of any
network server installation or operation. With the entire operation
of a corporation often dependent on the network and its associated
network servers, the loss of either could result in the failure of the
corporation. The details of disaster recovery are covered in
Chapter 8, “7.0—Disaster Recovery.”
One of the most innovative additions to disaster-recovery technology is
Hewlett-Packard’s One Button Disaster Recovery (OBDR). This tech-
nology allows a tape drive equipped with OBDR to restore the con-
tents of the network server without going through the usual restoration
steps. Using OBDR, there is no need to wipe the disk drives clean,
no need to reinstall a clean version of the network operating system, no
need to install service patches to the network operating system, and
no need to install the tape backup software before beginning a restora-
tion of the network server’s data from tape.
With OBDR, you just perform a full tape restore from the backup
tapes. This will restore the entire contents of the network server to
the state of its last full backup.
For more information on OBDR, see Hewlett-Packard’s web site
at www.products.storage.hp.com/eprise/main/storage/
tapebackup/selectionPages/obdr.htm.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 104
MISCELLANEOUS NETWORK
SERVER HARDWARE
. 0.17 Identify miscellaneous hardware components that are part
of a network server. Content may include the following:
• Video controller
• Video monitor
• Universal Serial Bus (USB)
• FireWire (IEEE 1394)
• Keyboard
• Mouse
• CD-ROM drive
• DVD-ROM drive
• Floppy disk drive
• Modem
Several miscellaneous hardware components can be included in a net-
work server. The following sections describe these various components.
Video Controller
A video controller doesn’t need to support high video resolutions on
the network server. A video controller that can support 1024 × 768 res-
olution and 64K colors should be sufficient for most network servers.
Video Monitor
A video monitor for a network server should support 800 × 600 or
better video resolution. A large video monitor (21 inches) might be
useful on a network server. An LCD display could also be used as a
monitor for a network server. If the video monitor is going to be
installed in a server rack, the video monitor must be physically small
enough to fit into the rack. The maximum size for a video monitor
that is going into a server rack is about 15 inches.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 105
TIP
server rack, then using a keyboard/video/mouse (KVM) switch will a keyboard/video/mouse (KVM)
switch.
allow one monitor (keyboard and mouse) to be switched among sev-
EXAM
eral network servers. This will save valuable space.
The IEEE 1394 standard (known as FireWire by Apple Computer) is for IEEE 1394 (FireWire).
an external serial bus that uses the SCSI-3 standards for communica-
EXAM
Keyboard
Every network server needs a keyboard for data input. The keyboard
can use a traditional round plug or a USB connector.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 106
Mouse
Every network server needs a mouse for the same reasons that a
desktop computer system needs a mouse: to interact with a graphical
user interface. In some cases, the mouse will be replaced with a dif-
ferent pointing device, such as a touch pad or a trackball.
CD-ROM Drive
A CD-ROM drive is required on a network server for operating
system and other software installations.
DVD-ROM Drive
A DVD-ROM drive on a network server can perform the same
functions as a CD-ROM drive. In addition, a DVD-ROM drive can
be useful on a video server to transfer video from the DVD to the
disk drives on the video server.
Modem
A modem or multiple modems might need to be installed in the
network server due to requirements of the role it will play in the net-
work. For example, a remote access server would need one or more
modems to support dial-in access to the network.
To implement a fax server, one or more fax modems (which differ
from data modems) must be installed in the network server.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 107
RACKING SERVERS
. 0.18 Identify situations when network servers should be
installed in a server rack and know the proper procedures
for installation of the network server into a server rack.
Identify specialized components and installation require-
ments for rack-mounted network servers. Content may
include the following:
• EIA rack unit (1.75 inches)
• Keyboard/video/monitor (KVM) switch
• Cable management
• Rack layout
• Security
• Cooling
• Rack-mounted keyboard/trackball
• Rack-mounted monitor/LCD panel
• Rack-mounted uninterruptible power supply (UPS)
To conserve space in the server room, network servers can be pur-
chased in a chassis that can be installed into a standard 19-inch
equipment rack. Figure 1.11 shows a diagram of a network server
rack.
Keyboard/Video/Monitor Switch
A keyboard/video/monitor (KVM) switch is a common component in
a network server rack. With multiple network servers installed in a
single equipment rack, it makes sense to have a single keyboard,
video display, and mouse that can be switched among the network
servers. A rack-mountable KVM switch usually fits within 1 rack
unit and provides the capability to share the keyboard, video display,
and mouse among the network servers in the rack.
Cable Management
With several network servers and other devices in a rack, the manage-
ment of the various cables becomes critical. Also, generally devices
that are installed in a rack can be pulled out of the rack for service,
much like a drawer can be pulled out from a cabinet. Pulling the net-
work server out of the rack without disconnecting all the cables run-
ning to the network server (LAN, power, video, mouse, keyboard,
and so on) is accomplished by hinged cable management arms
attached to the network server on one end and the rack on the other.
Rack Layout
The layout of the components in the rack is important. Generally the
rack-mounted UPSs are installed on the bottom of the rack due to
their weight. A rack-mounted keyboard should be at the proper height
to allow comfortable typing. The rack-mounted display adapter should
be at eye level. Most manufacturers of network server racks provide
software to aid in creating a layout of the devices in the network rack.
Security
The network server should have side panels (which are usually
optional) and front and rear doors that are lockable to provide
security for the network devices installed in the rack.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 109
Cooling
The network server rack should provide adequate ventilation to keep
the network devices in the rack as cool as possible. Most network
servers that are designed to be rack-mounted are designed to have a
flow of air from the front of the rack to the back. The front and rear
doors of the rack are often perforated to provide ventilation.
Rack-Mounted Keyboard/Trackball
Rack-mounted keyboards or keyboard/trackball combinations that
fit into a drawer that slides into the rack are available. This one key-
board can be attached to a rack-mounted KVM switch so that it can
control all the network servers in the rack. The keyboard drawer is
usually 2 rack units high.
Rack-Mounted UPS
Special UPSs designed to fit into a rack are available. These UPSs are
generally installed in the bottom of the rack because of their weight.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 110
R E V I E W B R E A K
This section discussed putting network servers into racks specifically
designed to conserve floor space by allowing multiple network
servers to be installed into a single rack. However, this generally
requires prepurchase planning. The network servers, as well as other
peripherals, should be designed to be installed in a server rack. The
network server should be purchased in a rack-mount chassis. Rack-
mountable UPSs, external disk systems, and special rack-mountable
keyboards are also available. Most major server vendors have soft-
ware available that can be used to configure the components to be
installed in a server rack.
The one drawback to a server rack is that when it is fully filled it can
weight well over 1,000 pounds. To move the rack from one location to
another, the network server and its components must be removed
before the rack is moved and reinstalled after the rack has been moved.
Novell NetWare 4
Novell’s NetWare version 4 was the first version of NetWare to
incorporate Novell’s Directory Service (NDS). NetWare 4 uses the
IPX/SPX protocol natively, and supports a non-native version of the
TCP/IP protocol. NetWare 4 supports the NetWare File System
(NWFS). The maximum file size on NWFS is 4GB. NetWare 4
supports software-based RAID 1 (both mirroring and duplexing)
and RAID 5.
For more information on Netware 4.2, see Novell’s web site at
www.novell.com/products/netware4.
Novell NetWare 5
Novell’s NetWare version 5 supports the TCP/IP protocol natively.
NetWare 5 also supports two different file systems: the NetWare
File System (NWFS) and the Novell Storage Services (NSS) File
System. The Novell Storage Services File System supports file sizes
up to 8 terabytes.
For more information on NetWare 5.1, see Novell’s web site at
www.novell.com/products/netware/.
UNIX
NOTE
Linux
Linux is a clone of the UNIX operating system initially created by
Linus Torvalds in 1991. The most attractive features of Linux are
that it is free and the source code is available (known as open source).
Many major hardware and software vendors are supporting Linux.
Examples of available Linux distributions include the following:
á Caldera Linux
á Red Hat Linux
á SuSE Linux
á VA Linux
R E V I E W B R E A K
This section described some of the various network operating sys-
tems that can be installed on an Intel-powered network server. The
answer to the question, “which network operating system is best?”
depends on the role the network server will perform, the operating
systems of existing network servers, network connectivity issues,
applications compatibility, and cost.
Antivirus Software
There is an even greater need to protect a network server from com-
puter viruses than to protect desktop computers from computer
viruses. When a virus gets onto your network server, it is more easily
spread to all the computers that access the network server. If the
computer virus is destructive, you could easily lose all the data stored
on the network server. Antivirus software should be near the top of
the list, along with backup software and the network server operating
system, to acquire as part of installing a new network server. Running
a network server without antivirus software is just not prudent.
Antivirus software for a network server should be configured to
update the virus signature file frequently, as new viruses are found
on practically a daily basis.
Documentation
Documentation is very important to the successful installation
and operation of a network server. The documentation supplied
with the network server should be read and stored in a safe place.
The installer of the network server should create a logbook for the
network that initially will contain the configuration information
for the network server. As the network server is upgraded and main-
tained, additional entries should be recorded in the logbook. A well-
maintained logbook is an invaluable tool when it is time to upgrade
the network server or when it is time to troubleshoot a problem
with the network server.
R E V I E W B R E A K
This section discussed the various hardware and software utilities
available to aid in the management of a network server. Using the net-
work server setup, installation, and configuration software provided by
the network server vendor can make the job of setting up the network
server much easier. This software also sets up the diagnostic partition,
special hardware drivers, and it handles RAID system configuration.
Special network server management hardware and software that is
available from the network server vendor is a wise investment that
quickly recovers its own cost, especially if multiple network servers
need to be managed together.
continues
06 Serv+ Ch_01 7/10/01 9:41 AM Page 118
continued
Running a network server without antivirus software is just asking
for trouble. (I consider it to be an absolute requirement.)
Be sure to document everything about the setup, installation, config-
uration, upgrading, and maintenance of the network server in a log-
book. It will prove to be invaluable later, when troubleshooting a
network server problem.
FIGURE 1.12
Network Server Installation Plan Worksheet
Network Server Installation Plan Worksheet.
Server Role ________________________________________________________________
Other __________________________________________________________________
1 ___________________________________________________________________
2 ___________________________________________________________________
3 ___________________________________________________________________
4 ___________________________________________________________________
FIGURE 1.13
Server Memory. Server Memory
Memory Type
/__/ EDO
/__/ EDO Buffered
/__/ SDRAM
/__/ SDRAM Registered
/__/ RDRAM
/__/ Other ________________________
FIGURE 1.14
ATA, ATA2, Ultra ATA Disk Controller ATA, ATA2, Ultra ATA Disk Controller.
Complete one copy of this page for each SCSI Bus Controller in the Network Server
FIGURE 1.15
SCSI Bus Controller.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 122
FIGURE 1.16
RAID Controller. RAID Controller
Complete one copy of this page for each RAID Controller in the Network Server
Channel 1, SCSI ID 0
Channel 1, SCSI ID 1
Channel 1, SCSI ID 2
Channel 1, SCSI ID 3
Channel 1, SCSI ID 4
Channel 1, SCSI ID 5
Channel 1, SCSI ID 6
Channel 1, SCSI ID 7
Channel 1, SCSI ID 8
Channel 1, SCSI ID 9
Channel 1, SCSI ID 10
Channel 1, SCSI ID 11
Channel 1, SCSI ID 12
Channel 1, SCSI ID 13
Channel 1, SCSI ID 14
Channel 1, SCSI ID 15
Channel 1, SCSI ID 0
Channel 1, SCSI ID 1
Channel 1, SCSI ID 2
Channel 1, SCSI ID 3
Channel 1, SCSI ID 4
Channel 1, SCSI ID 5
Channel 1, SCSI ID 6
Channel 1, SCSI ID 7
Channel 1, SCSI ID 8
Channel 1, SCSI ID 9
Channel 1, SCSI ID 10
Channel 1, SCSI ID 11
Channel 1, SCSI ID 12
Channel 1, SCSI ID 13
Channel 1, SCSI ID 14
Channel 1, SCSI ID 15
06 Serv+ Ch_01 7/10/01 9:41 AM Page 123
FIGURE 1.17
PCI Bus Layout PCI Bus Layout.
PCI Bus# Slot# Adapter Installed
Other Buses
Is there a working network connection available for the new server (yes/no)?
Do the electrical receptacles match the electrical plugs on the UPS (yes/no)?
Does the installation site have enough cooling to support the new server (yes/no)?
FIGURE 1.18
Server Installation Site Preparation.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 124
FIGURE 1.19
Completed Network Server Installation Plan Network Server Installation Plan Worksheet
Worksheet. Server Role Web Server using Microsoft Internet Information Server
Server Processor(s) (Quantity, Brand, Model, Speed, L2 Cache) 2, Intel, Pentium III Xeon,
700MHz, 256KB
Other
Backup Software (Brand, Version) Mighty Mite Backup Software Version 2.x
Video Adapter (Brand, Model, Video RAM, Maximum Resolution) Ultra VGA Computer
Model 33, 8MG, 2024 x 768
Uninterruptible Power Supply (Brand, Model, VA rating) Big Battery Backup, Model 12x,
2000 VA
2.
3.
4.
Other Adapters
06 Serv+ Ch_01 7/10/01 9:41 AM Page 125
FIGURE 1.20
Server Memory
Completed Server Memory.
Memory Form Factor
/__/ SIMM
/X/ DIMM
/__/ RIMM
/__/ Other
Memory Type
/__/EDO
/__/EDO Buffered
/X/ SDRAM
/__/SDRAM Registered
/__/SDRAM
/__/Other
Memory Error Checking
/__/ none
/__/ Parity
/X/ ECC
Memory Module Connectors
/__/ Tin
/X/ Gold
FIGURE 1.21
Completed ATA, ATA2, Ultra ATA Disk Controller. ATA, ATA2, Ultra ATA Disk Controller
1. Disk Controller Brand and Model
Complete one copy of this page for each SCSI Bus Controller in the Network Server
SCSI Bus Controller (Brand, Model, SCSI Standard) Builtin, One, SCSI-2
Channel 1, SCSI ID 0
Channel 1, SCSI ID 1
Channel 1, SCSI ID 2
Channel 1, SCSI ID 3
Channel 1, SCSI ID 4
Channel 1, SCSI ID 5 Brand 2, Model A Tape Backup
Channel 1, SCSI ID 6
Channel 1, SCSI ID 7
Channel 1, SCSI ID 8
Channel 1, SCSI ID 9
Channel 1, SCSI ID 10
Channel 1, SCSI ID 11
Channel 1, SCSI ID 12
Channel 1, SCSI ID 13
Channel 1, SCSI ID 14
Channel 1, SCSI ID 15
Channel 2, SCSI ID 0
Channel 2, SCSI ID 1
Channel 2, SCSI ID 2
Channel 2, SCSI ID 3
Channel 2, SCSI ID 4
Channel 2, SCSI ID 5
Channel 2, SCSI ID 6
Channel 2, SCSI ID 7
Channel 2, SCSI ID 8
Channel 2, SCSI ID 9
Channel 2, SCSI ID 10
Channel 2, SCSI ID 11
Channel 2, SCSI ID 12
Channel 2, SCSI ID 13
Channel 2, SCSI ID 14
Channel 2, SCSI ID 15
FIGURE 1.22
Completed SCSI Bus Controller.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 127
Channel 2, SCSI ID 0
Channel 2, SCSI ID 1
Channel 2, SCSI ID 2
Channel 2, SCSI ID 3
Channel 2, SCSI ID 4
Channel 2, SCSI ID 5
Channel 2, SCSI ID 6
Channel 2, SCSI ID 7
Channel 2, SCSI ID 8
Channel 2, SCSI ID 9
Channel 2, SCSI ID 10
Channel 2, SCSI ID 11
Channel 2, SCSI ID 12
Channel 2, SCSI ID 13
Channel 2, SCSI ID 14
Channel 2, SCSI ID 15
06 Serv+ Ch_01 7/10/01 9:41 AM Page 128
FIGURE 1.24
Completed PCI Bus Layout. PCI Bus Layout
PCI Bus# Slot# Adapter Installed
Other Buses
Is there a working network connection available for the new server (yes/no)? YES
Do the electrical receptacles match the electrical plugs on the UPS (yes/no)? YES
Does the installation site have enough cooling to support the new server (yes/no)? YES
FIGURE 1.25
Completed Server Installation Site Preparation.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 129
CHAPTER SUMMARY
In this chapter, you learned about the technologies used to create
KEY TERMS
network servers. The differences between the two major types of
networks—peer to peer and server centric—were identified. The • 10BASE-2
major categories of network server, based on size, were described. • 10BASE-5
The major network server roles were described in detail.
• 10BASE-F
The major components of a network server were then discussed to
give the server hardware specialist and the network engineer the • 10BASE-T
insight necessary to select components for the network server. The • 100BASE-TX
goal is to select the proper network server components so that it
• 802.3
can adequately fulfill its role in the network. Those components
fit into major categories such as processors, memory, the network • 802.5
subsystem, and the disk subsystem.
• 802.11
Miscellaneous additional topics related to a network server, includ-
• Adapter teaming
ing the network server operating system, management software, tape
backup system, uninterruptible power supplies, and network server • Adapter load balancing
racks, were covered to fill in any gaps related to network server selec- • Adaptive fault tolerance
tion. Finally, the creation of an installation plan was discussed.
• Advanced Intelligent Tape
• American National Standards
Institute (ANSI)
• American Standard Code for
Information Interchange (ASCII)
• AT Attachment (ATA)
• Berkeley Internet Name Domain
(BIND)
• Buffered memory
• Cache
• Daemon
• Departmental server
• Desktop Management Interface
(DMI)
06 Serv+ Ch_01 7/10/01 9:41 AM Page 130
CHAPTER SUMMARY
CHAPTER SUMMARY
CHAPTER SUMMARY
A P P LY Y O U R K N O W L E D G E
Exercises B. At least 100GB of disk storage to hold the
200,000 distinct web pages and their associated
1.1 Create an Installation Plan graphic images. Plan for fault tolerance of
the disk storage by using either hardware or
This exercise is intended to give you some hands-on software RAID.
experience planning the specifications for a network
server installation. You can repeat it for different server C. Plan for high availability by using fault-tolerant
vendors and for different server roles. components and redundancy when necessary.
Estimated time: One to four hours 4. Make sure that the installation site has adequate
power, cooling, and room to install the network
1. Print a copy of the 01 Installation server.
Plan Worksheet.doc file that is
on the CD-ROM. 5. Don’t forget the tape backup system, and the
UPS.
2. Using the “Installation Plan Worksheet,” plan the
configuration and installation of a network server 6. Visit a network server vendor web site (Compaq,
that meets the following specifications. Dell, IBM, HP) and select and configure a net-
work server model offered by that vendor that
A. The network server will be a web server. meets the specifications.
B. The web server will run many CGI or ASP Repeat step 6 using a different network server vendor
scripts. than the one originally selected.
C. The web server will contain approximately As noted above, this entire exercise can be repeated for
200,000 distinct web pages, many of which any network server role.
utilize lots of graphic images.
D. The web server will run Microsoft Windows
2000 Server and it will use Microsoft Internet
Review Questions
Information Server (IIS) as the web server. 1. What is the purpose of a NAT server?
3. Based on these specifications, the web server will 2. What is the purpose of an uninterruptible power
probably need the following: supply?
A. At least 256MB of memory; probably 3. What is the purpose of a KVM switch?
512MB is a better choice. Windows 2000
4. What is sent to a Wake-On-LAN Network
Server needs at least 128MB of memory. The
Interface Card to cause it to power up the
web server software uses the remaining mem-
computer?
ory in the network server to cache frequently
accessed web pages. The CGI and ASP scripts 5. What is SMP?
will use memory to run.
6. What is adapter teaming?
06 Serv+ Ch_01 7/10/01 9:41 AM Page 134
A P P LY Y O U R K N O W L E D G E
7. How many devices can be attached to an IEEE 25. What is the data transfer speed of FireWire?
1394 bus?
8. What is the minimum number of disk drives
required to implement RAID 0/1? Exam Questions
9. What is the purpose of a DNS server? 1. Which RAID level provides improved disk
10. What is the name of the system bus created by input/output, but provides no redundancy?
IBM for the IBM PS/2 microcomputer? A. RAID 0
11. What is the purpose of I2O technology? B. RAID 1
12. What are the common form factors for network C. RAID 3
server memory?
D. RAID 5
13. What is the speed of Fast Ethernet?
2. Which RAID level provides redundancy at the
14. What is the difference between write thru and expense of the loss of 50% of the disk storage
write back cache? capacity?
15. What is the purpose of S.M.A.R.T technology? A. RAID 0
16. What is the data transfer rate of USB? B. RAID 1
17. What is the purpose of racking servers? C. RAID 3
18. Why is ECC memory important to network D. RAID 5
server?
3. Which of the following RAID levels provide fault
19. What is the downside to using disk mirroring? tolerance using parity information and a mini-
20. What is the native storage capacity of a Digital mum of three disk drives?
Audio Tape (DAT) using the DDS-4 format? i. RAID 0
21. What are the reasons for having more than one ii. RAID 1
NIC in a network server?
iii. RAID 4
22. Which network server operating systems support
the NTFS file system? iv. RAID 5
A P P LY Y O U R K N O W L E D G E
4. Which of the following network server roles can 7. Which of the following network operating
be used to attach a corporate LAN to the systems can utilize the NTFS file system?
Internet?
i. Novell NetWare
i. Router
ii. Microsoft Windows NT Server
ii. Firewall
iii. Microsoft Windows 2000 Server
iii. Proxy server
iv. Linux
iv. Network Address Translation (NAT) server
A. i and iv
A. i and ii
B. i, ii, and iv
B. i, ii, and iii
C. ii and iii
C. ii and iv
D. iv only
D. i, ii, iii, and iv
8. Which of the following memory technologies can
5. Which of the following buses are “local buses”? be buffered?
i. PCI A. EDO
ii. ISA B. SDRAM
iii. AGP C. DDR DRAM
iv. EISA D. RDRAM
A. ii and iv 9. You need to install 80 disk drives in an external
enclosure and control them from a single
B. i and iii
controller in the network server. Which disk
C. i and ii technology will you use?
D. iii and iv A. SCSI-1
6. What is the native (not compressed) storage B. Ultra 100
capacity of a DAT24 tape drive?
C. FC-AL
A. 4GB
D. SCSI-3
B. 8GB
C. 12GB
D. 40GB
06 Serv+ Ch_01 7/10/01 9:41 AM Page 136
A P P LY Y O U R K N O W L E D G E
10. Which network device supports port aggregation? 14. Which of the following network services is used
to automatically configure TCP/IP?
A. Hub
A. DNS
B. Switch
B. WINS
C. Bridge
C. DHCP
D. Router
D. SAN
11. Which SCSI standard defines standards for both
parallel and serial communications? 15. What is the data transfer speed of Ultra2 SCSI?
A. SCSI-0 A. 10MBps
B. SCSI-1 B. 20MBps
C. SCSI-2 C. 40MBps
D. SCSI-3 D. 80MBps
12. What is the maximum bus (cable) length for a 16. Which of the following has a data transfer rate
SCSI system that uses Low Voltage Differential of 12Mbps and allows attachment of up to
(LVD) signaling? 127 Plug and Play devices?
A. 3 meters A. FireWire
B. 6 meters B. USB
C. 12 meters C. Serial ATA
D. 25 meters D. Serial Storage Architecture
13. Which of the following TCP/IP protocols can be 17. What type of memory is used in the Level 2
used to manage network devices? cache of an Intel processor?
A. SMTP A. DRAM
B. FTP B. SRAM
C. SNMP C. SDRAM
D. SNA D. ECC
06 Serv+ Ch_01 7/10/01 9:41 AM Page 137
A P P LY Y O U R K N O W L E D G E
18. Which of the following RAID systems provide C. Intel
disk fault tolerance by duplicating data in two
D. Compaq
locations?
22. Server clustering can be achieved using which of
A. RAID 0
the following techniques?
B. RAID 1
i. Mirrored disk drives
C. RAID 5
ii. Shared disk drives
D. RAID 3
iii. Shared nothing
19. What is the name of the magnetic tape technol-
iv. Redundant NICs
ogy that is an enhanced version of Quarter Inch
Cartridge (QIC) tape technology? A. i and iv
A. Enchanced QIC B. ii and iv
B. AIT C. i, ii, and iii
C. Travan D. iv only
D. DLT 23. Which of the following high-speed technologies
can be used to create a storage area network (SAN)?
20. Which of the following memory technologies
operates at the clock speed of the memory bus? i. Fibre Channel
i. EDO DRAM ii. Serial Storage Architecture
ii. SDRAM iii. Universal Serial Bus
iii. DDR SDRAM iv. FireWire
iv. Direct Rambus A. i and ii
A. i and ii B. i and iii
B. ii and iii C. ii and iii
C. iii and iv D. iii and iv
D. i and iv 24. What is the maximum cable length for a
ATA/IDE disk channel?
21. What company created the PCI bus standard?
A. 12 inches
A. Microsoft
B. 18 inches
B. Kingston
C. 1 meter
D. 3 meters
06 Serv+ Ch_01 7/10/01 9:41 AM Page 138
A P P LY Y O U R K N O W L E D G E
25. What is the bus speed of the PCI 2.1 specification? 4. A magic packet, which is a special TCP/IP packet
that contains the IP addresses of the computer to
A. 33MHz
be awakened, repeats 16 times in the data portion
B. 66MHz of the TCP/IP packet. For more information, see
C. 132MHz the section “Wake-On-LAN.”
A P P LY Y O U R K N O W L E D G E
12. The common form factors for memory in net- 19. Disk mirroring has a data storage loss of 50%
work servers are SIMMs (single in-line memory and both disk drives in a mirrored set are
module), DIMMs (dual in-line memory module) attached to a single disk controller. Failure of the
and RIMMs. For more information, see the sec- disk controller means loss of the mirror set. For
tion “Memory Form Factors.” more information, see the section “RAID 1.”
13. Fast Ethernet operates at 100Mbps. For more 20. The native storage capacity of a DAT tape
information, see the section “Fast Ethernet.” using DDS-4 format is 20GB. For more infor-
mation, see the section“Digital Audio Tape.”
14. The difference between write thru and write back
cache is that with write thru cache data is written 21. Multiple NICs in a network server allow the net-
to main memory and the cache at the same time. work server to be attached to more than one
In write back cache, the data is written only to physical network. More importantly, multiple
the cache, which writes it to main memory when NICs allow for adapter teaming. For more infor-
necessary. For more information, see the sections mation, see the section “Adapter Teaming.”
“Write Thru Cache” and “Write Back Cache.”
22. The NTFS file system is supported by Microsoft
15. The Self-Monitoring, Analysis and Reporting Windows NT Server 4 and Microsoft Windows
Technology (S.M.A.R.T) is a disk drive technol- 2000 Server. For more information, see the sec-
ogy that monitors disk drive parameters for indi- tion “Network Operating Systems.”
cations of failure. For more information, see the
23. Common frontside bus speeds are 66MHz,
section “Self-Monitoring, Analysis and Reporting
100MHz, and 133MHz. For more information,
Technology (S.M.A.R.T).”
see the section “Memory Bus.”
16. The data transfer rate of a Universal Serial Bus
24. A WINS server is used to register a computer’s
(USB) is 12Mbps. For more information, see the
NetBIOS name and IP address when it is pow-
section “Universal Serial Bus.”
ered up. The WINS server is also used to resolve
17. Putting network servers into a rack allows for the a NetBIOS name to an IP address. For more
installation of many servers in a small amount of information, see the section “Windows Internet
floor space. For more information, see the section Naming Service Server.”
“Racking Servers.”
25. The data transfer speed of IEEE 1394, aka
18. Error Correcting Code (ECC) memory can FireWire, is 400Mbps. For more information, see
detect and correct single-bit errors and detect the section “FireWire (IEEE 1394).”
double-bit errors in the memory of network
servers. For more information, see the section
“Memory Error Checking.”
06 Serv+ Ch_01 7/10/01 9:41 AM Page 140
A P P LY Y O U R K N O W L E D G E
Answers to Exam Questions 8. A. Only EDO memory can be buffered. For
more information, see the section “Buffered
1. A. RAID 0 provides improved disk input/output Memory and Registered Memory.”
but does not provide fault tolerance. For more
information, see the section “RAID 0.” 9. C. Only FC-AL (Fibre Channel Arbitrated Loop)
of the choices in the answer has the capacity to
2. B. RAID 1 (disk mirroring or disk duplexing) attach 80 disk drives to a single disk controller.
provides redundancy by writing the same data to For more information, see the section “Fibre
two disk drives, but at the cost of 50% of the Channel Arbitrated Loop.”
storage capacity of the disk drives being mirrored
or duplexed. For more information, see the sec- 10. B. A switch can support port aggregation. For
tion “RAID 1.” more information, see the section “Port
Aggregation.”
3. B. RAID 4 and RAID 5 both require a mini-
mum of three disk drives and provide fault toler- 11. D. SCSI-3 has standards for both parallel and
ance by means of parity. For more information, serial communications. For more information,
see the sections “RAID 4” and “RAID 5.” see the section “SCSI-3”.
4. D. A router, a firewall, a proxy server, and a NAT 12. C. The maximum cable length for Low Voltage
server can all be used to connect a corporate Differential (LVD) SCSI is 12 meters. For more
LAN to the Internet. Each server has a different information, see Table 1.1.
purpose, but they all connect a corporate LAN to 13. C. The Simple Network Management Protocol
the Internet. For more information, see the sec- (SNMP) can be used to manage network devices.
tions “Router,” “Firewall,” “Proxy Server,” and For more information, see the section “Simple
“Network Address Translation Server.” Network Management Protocol.”
5. B. Both PCI and AGP are local buses. For more 14. C. Dynamic Host Configuration Protocol
information, see the sections “Peripheral (DHCP) can be used to automatically configure
Component Interconnect” and “Accelerated TCP/IP clients. For more information, see
Graphics Port.” the section “Dynamic Host Configuration
6. C. The native storage capacity of DAT24 is 12GB. Protocol Server.”
The compressed storage capacity of DAT24 is 15. B. Ultra2 SCSI can transfer data at 20MBps. For
24GB. For more information, see Table 1.9. more information, see Table 1.1.
7. C. Both Microsoft Windows NT Server and 16. B. The Universal Serial Bus (USB) has a data
Microsoft Windows 2000 Server can use the transfer speed of 12Mbps and supports up to
NTFS file system. For more information, see the 127 Plug and Play devices. For more information,
sections “Microsoft Windows NT Server” and see the section “Universal Serial Bus.”
“Microsoft Windows 2000 Server.”
17. B. Static RAM (SRAM) is used in the Level 2
cache in Intel processors. For more information,
see the section “Processor Cache.”
06 Serv+ Ch_01 7/10/01 9:41 AM Page 141
A P P LY Y O U R K N O W L E D G E
18. B. RAID 1 (disk mirroring or disk duplexing) 22. C. Server clusters can be created using any of
provides disk fault tolerance by duplicating data these three methods: mirrored disk drives, shared
to two different locations. For more information, disk drives, and shared nothing. For more infor-
see the section “RAID 1.” mation, see the section “Server Clustering.”
19. C. The Travan tape technology is an enhanced 23. A. Two technologies that can be used to create a
version of the QIC tape technology. It does pro- storage area network (SAN) are Fibre Channel and
vide some backward compatibility with QIC Serial Storage Architecture. For more information,
tapes. For more information, see the section see the section “Storage Area Network.”
“Travan Cartridge Tape.”
24. B. The maximum cable length for an ATA/IDE
20. B. Both Synchronous DRAM (SDRAM) and dou- channel is 18 inches. For more information, see
ble data rate (DDR) synchronous DRAM operate the section “Integrated Drive Electronics.”
at speed of the memory bus. For more informa-
25. B. The PCI 2.1 specification has a bus speed of
tion, see the section “Memory Technologies.”
66MHz. For more information, see the section
21. C. The PCI bus standard was created by Intel “Peripheral Component Interface.”
Corporation. For more information, see the
section “Peripheral Component Interface.”
1. Albitz, Paul and Cricket Liu. DNS and BIND, 6. Microsoft Windows NT 4.0 Server
4th Edition. O’Reilly & Associates (2001). www.microsoft.com/catalog/
ISBN 0-5960-0158-4. display.asp?site=427&subid=22&pg=1
2. Field, Gary and Peter M. Ridge. The Book of 7. Microsoft Exchange Server
SCSI: I/O for the New Millennium. No Starch www.microsoft.com/exchange/default.htm
Press (2000). ISBN 1-8864-1110-7.
8. Microsoft SQL Server
3. Mueller, Scott. Upgrading and Repairing PCs, www.microsoft.com/sql/default.htm
12th Edition. Que Publishing (2000). ISBN
9. Microsoft Host Integration Server
0-7897-2303-4.
www.microsoft.com/hiserver/
4. Spurgeon, Charles E. Ethernet: The Definitive
10. Microsoft SNA Server
Guide. O’Reilly & Associates (2000). ISBN
www.microsoft.com/sna/default.asp
1-5659-2660-9.
11. Microsoft Proxy Server
5. Dedek, Jan and Gary Stephens. What Is Fibre
www.microsoft.com/proxy/ and
Channel?, 5th Edition. Ancot Corporation
www.microsoft.com/isaserver/
(2000) ISBN 0-9637-4396-1.
06 Serv+ Ch_01 7/10/01 9:41 AM Page 142
A P P LY Y O U R K N O W L E D G E
Suggested Readings and Resources
21. Novell Netware Cluster Services for 32. Desktop Management Task Force on DMI
NetWare 5.1 www.dmtf.org/spec/dmis.html
www.novell.com/products/clusters/ncs/
33. Charles Spurgeon’s Ethernet IEEE 802.3 Web
22. IBM OS/2 Warp Server site www.ots.utexas.edu/ethernet/
http://www-4.ibm.com/
34. 1394 Trade Association www.1394ta.org/
software/os/warp/warp-server/
35. Kingston’s Ultimate Memory Guide
www.kingston.com/tools/umg/default.asp
07 Serv+ Ch_02 7/11/01 10:00 AM Page 143
OBJECTIVES
OBJECTIVES OUTLINE
OUTLINE
Match Processor Stepping (Version) 163 Backup System Hardware Installation 178
Buses 163 Uninterruptible Power-Supply Installation 178
Peer Buses 163 Miscellaneous Server Components
Bridged Buses 164 Installation 179
Memory Installation 164 Monitor 179
Match Memory Type 165 Keyboard 179
Match Memory Speed 165 Mouse 179
Disk Subsystem Installation 166 Video Adapter 181
IDE/ATA Disks 166 Server Management Adapter 181
EIDE/ATA-2 Disks 167 CD-ROM/DVD-ROM Drive 181
Ultra ATA/Ultra DMA 168 Redundant Component Installation 181
SCSI Disks 168 Power Supply 181
SCSI Disk Types 170 Cooling Fan 182
SCSI-1 170 Redundant NIC 182
SCSI-2 171 Server-Rack Installation 182
SCSI-3 173 Power-On Test 184
Installing SCSI 173 Document, Document, Document 185
Fibre Channel Arbitrated Loop 175
Network Subsystem Installation 175
Chapter Summary 186
Ethernet 175
Token Ring 176
Network Interface Cards 176
Apply Your Knowledge 188
Network Cabling 176
NIC to Hub 177
Hub to Hub 177
07 Serv+ Ch_02 7/10/01 9:48 AM Page 146
S T U DY S T R AT E G I E S
To prepare for the Installation objectives of the . Use the ExamGear test engine on the CD-
Server+ exam, do the following: ROM that accompanies this book for addi-
. Read the objectives at the beginning of this tional exam questions concerning this
chapter. material.
. Study the information in this chapter. . Review the Test Tips available in various places
in this chapter. Make certain that you under-
. Review the objectives again. stand the information in the Test Tip. If you
. Answer the Review and Exam Questions at the don’t fully understand the Test Tip, review the
end of the chapter and check your results. information in the chapter related to this topic.
07 Serv+ Ch_02 7/10/01 9:48 AM Page 147
INTRODUCTION
The Installation section covers the pre-installation and installation of
network server hardware. This section is the topic of approximately
17% of the exam questions.
This chapter covers the two objectives of the Installation job dimen-
sion. The first objective deals with reviewing the installation plan,
checking all the network server hardware, and checking that the instal-
lation site is ready for the network server to be installed. The second
objective covers the actual installation of the network server hardware.
This objective ends with an actual power-on of the network server to
see that all the network server hardware is recognized after installation.
1.
2.
3.
4.
1.
2.
3.
4.
Site Checklist
Is space available to install the new network server?
Is there a working network connection available for the new network server?
Does the installation site have enough cooling to support the new network server?
FIGURE 2.3
IDE/ATA/EIDE/ATA2 Disk Controller
IDE/ATA/EIDE/ATA2 Disk Controller.
1. Disk Controller Brand Model
Channel 1,SCSI ID 0
Channel 1, SCSI ID 1
Channel 1,SCSI ID 2
Channel 1,SCSI ID 3
Channel 1,SCSI ID 4
Channel 1,SCSI ID 5
Channel 1,SCSI ID 6
Channel 1,SCSI ID 7
Channel 1,SCSI ID 8
Channel 1,SCSI ID 9
Channel 1,SCSI ID 10
Channel 1,SCSI ID 11
Channel 1, SCSI ID 12
Channel 1,SCSI ID 13
Channel 1,SCSI ID 14
Channel 1,SCSI ID 15
Channel 2,SCSI ID 0
Channel 2,SCSI ID 1
Channel 2,SCSI ID 2
Channel 2,SCSI ID 3
Channel 2,SCSI ID 4
Channel 2,SCSI ID 5
Channel 2,SCSI ID 6
Channel 2,SCSI ID 7
Channel 2,SCSI ID 8
Channel 2,SCSI ID 9
Channel 2,SCSI ID 10
Channel 2,SCSI ID 11
Channel 2,SCSI ID 12
Channel 2,SCSI ID 13
Channel 2, SCSI ID 14
Channel 2,SCSI ID 15 FIGURE 2.4
SCSI Bus Controllers.
07 Serv+ Ch_02 7/10/01 9:48 AM Page 152
FIGURE 2.5
RAID Controllers. RAID Controllers
Complete one copy of this page for each RAID Controller in the Network Server
Channel 1,SCSI ID 0
Channel 1, SCSI ID 1
Channel 1,SCSI ID 2
Channel 1,SCSI ID 3
Channel 1,SCSI ID 4
Channel 1,SCSI ID 5
Channel 1,SCSI ID 6
Channel 1,SCSI ID 7
Channel 1,SCSI ID 8
Channel 1,SCSI ID 9
Channel 1,SCSI ID 10
Channel 1,SCSI ID 11
Channel 1, SCSI ID 12
Channel 1,SCSI ID 13
Channel 1,SCSI ID 14
Channel 1,SCSI ID 15
Channel 2,SCSI ID 0
Channel 2,SCSI ID 1
Channel 2,SCSI ID 2
Channel 2,SCSI ID 3
Channel 2,SCSI ID 4
Channel 2,SCSI ID 5
Channel 2,SCSI ID 6
Channel 2,SCSI ID 7
Channel 2,SCSI ID 8
Channel 2,SCSI ID 9
Channel 2,SCSI ID 10
Channel 2,SCSI ID 11
Channel 2,SCSI ID 12
Channel 2,SCSI ID 13
Channel 2, SCSI ID 14
Channel 2,SCSI ID 15
Add to Tool Kit A server hardware addition of the network server to be installed. The installation plan
specialist’s tool kit should include a should state the heat output of the devices that will be part of the
digital thermometer. network server installation (chassis, monitor, UPS, and so on). It may
be difficult to ascertain the current heat load (measured in British
Thermal Units, BTUs) of the room where the network server is to be
installed. An adequate indicator of the available cooling capacity of
the room where the network server will be installed is the room tem-
perature. If the temperature of the room is more than 72 degrees
Fahrenheit without the new network server being installed, it is
doubtful that there is adequate cooling for the new network server.
Most network servers use internal fans for cooling. For these fans to
work properly and provide adequate cooling there has to be ade-
quate clearance around the server for proper airflow. The documen-
tation of the network server will specify the cooling clearances.
07 Serv+ Ch_02 7/10/01 9:48 AM Page 153
FIGURE 2.6
UPS Sizing Worksheet
UPS Sizing Worksheet.
This worksheet is designed to aid in the sizing of an uninterruptable power supply (UPS) to
meet the needs of a network server. UPSs are measured in terms of VA (volts-amps). List all
the equipment that will be protected by the UPS (that is, all equipment that will be plugged
into the UPS). Use Table 1 to list equipment that has voltage and amperage listed. Use Table 2
to list all the equipment that has power consumption listed in watts. Add all numbers in
the VA columns of both Table 1 and Table 2. The total VA rating is the minimum size to
select for the UPS that will support the network server.
In the following table, list all the equipment that has voltage and amperage ratings listed.
Multiply voltage times amperage to get the VA rating.
Table 1
Equipment Voltage (110/220) Amperage VA (Volts x Amps)
In the following table, list all the equipment that has wattage ratings listed. Multiply watts
times 1.5 to get the VA rating estimate.
Table 2
Equipment Wattage VA (Watts x 1.5)
FIGURE 2.7 Total VA requirements of all network server equipment to be attached to the UPS:
for a network server when they are ordered. The server hardware
NOTE
specialist must be able to take all the components and assemble Wide-Tower to Rack-Mount Conversion
them into a working network server. Some wide-tower server chassises can
be easily converted to a rack-mount
configuration with the purchase of a
Server Chassis special rack-mount adapter kit from
Verify that the server chassis is the correct model that was ordered the server manufacturer.
and the correct form factor. Most server chassises are either of a
tower configuration, a wide- or “fat-” tower configuration, or a rack-
mount configuration. Make sure that if you ordered model X in a
tower configuration, that indeed Model X in a tower configuration
is delivered, and not Model X in a rack-mount configuration (see
Figures 2.8, 2.9, and 2.10).
Server Rack
A rack-mount server chassis must be mounted in an equipment rack
designed for rack-mounted hardware. Most server vendors that sell
servers in a rack-mount chassis also sell the server rack. Server racks
generally come in several sizes (heights). Server rack size is measured
in rack units (U). A standard rack unit is 1.75 inches. A new net-
work server will either be installed in an existing server rack or a
new server rack will be purchased for the new network server. Server
racks generally have both front and rear doors and side panels to
form a completely enclosed unit. The doors and side panels help
form both a secure environment and also are designed to aid in
cooling the equipment in the rack by controlling airflow patterns. In
most cases, multiple server racks can be joined together to form an
installation space for a large number of servers. Figure 2.11 shows an
example of a server rack. FIGURE 2.8
Server tower chassis.
A rack-mounted server installation conserves space in the room
where the servers reside. If you go with a rack-mounted configura-
tion, however, generally all equipment must be designed to fit into a
NOTE
server rack. This includes the UPS, the keyboard/mouse, and the Do Not Move a Loaded Server Rack
monitor (either LCD or CRT) for the server. The equipment instal- Never attempt to move a server rack
that has equipment installed in it. A
lation in the server rack should be done to the server vendor’s
fully loaded server rack might well
specifications. Several server vendors offer software to aid in the
weigh in excess of 1,000 pounds.
proper placement of the equipment into the server rack. For exam- Remove all of the equipment first,
ple, all server vendors recommend that the UPS be installed on the then relocate the server rack, lastly
bottom of the rack. This is done for several reasons, including the reinstall the equipment removed from
weight of the rack-mount UPS and the need to avoid the possible the server rack.
damage to a network server caused by a leaking UPS battery.
07 Serv+ Ch_02 7/10/01 9:48 AM Page 156
Processor(s)
Network servers commonly have more than one processor. Verify
that the correct number and type of processors are available for the
network server. Some server vendors install all the processors that
were ordered; other server vendors supply the server with zero or one
processor installed, and the server hardware specialist must install
any additional processors. Verify that the processors are of the same
type, speed, and stepping (version). Also verify that each processor
has the same size of L2 cache. Follow the network server vendor’s
instructions for installing additional processors.
Memory
FIGURE 2.9
Server wide-tower chassis. Network servers generally require a considerable amount of memory
for them to adequately perform their function. Some server vendors
install all the memory that was ordered; however, other server ven-
dors supply the server with a “standard” amount of memory, and the
server hardware specialist must install the memory above the stan-
dard amount. Verify that the server has the amount of memory that
was ordered. If some of the memory must be installed, verify that
the memory is the correct type for the server and follow the server
hardware vendor’s instructions for installation of the additional
memory. This is a critical step. Some servers require that memory be
installed in groups of 2 or 4 memory modules. (Check the server
vendor’s documentation to see whether memory modules must be
installed in groups of 2 or 4, instead of singly.) Failure to install the
memory correctly will result in the server either not recognizing all
FIGURE 2.10 the memory installed or the server failing to start at all.
Server rack-mount chassis.
Disk Drives
Many network server configurations require that a large amount of
disk storage be available in the server. Verify that the disk drives are
the correct size, speed, and type (IDE/ATA, EIDE/ATA-2, SCSI,
SCSI-2, SCSI-3). The disk drives may come installed in the server
chassis or they might be shipped in separate cushioned boxes to
avoid damage. The installation plan will specify which disk drives
(size, speed, and type) should be installed in the network server.
07 Serv+ Ch_02 7/10/01 9:48 AM Page 157
Monitor
Verify that the monitor for the network server is as ordered. In
general the monitor should support VGA resolution of at least
1024 × 768 dots per inch. The monitor can be the traditional CRT
type or one of the newer LCD flat-panel monitors. If the monitor is
to be rack mounted, a special monitor shelf will need to be pur-
chased. Special rack-mounted LCD flat-panel monitors are available;
these often include a keyboard/trackball combination in a single
rack drawer.
Keyboard
Verify that there is a keyboard for the server. There are no special
requirements for the keyboard that will be used with a server. A
rack-mounted keyboard is available for rack-mounted server
installations. This rack-mounted keyboard often includes a built-in
trackball or touch pad to substitute for a mouse.
Mouse
Verify that there is a mouse for the server. There are no special require-
ments for the mouse that will be used with a server. The mouse might FIGURE 2.11
A server rack.
require a serial connection, a Universal Serial Bus (USB) connection,
or a PS/2-type mini-DIN connection. Verify that the type of connec-
tion required by the mouse is available on the network server. In a
rack-mounted configuration, the mouse is often replaced by a track-
ball or touch pad built in to the rack-mounted keyboard.
Backup System
Verify that the backup system is as specified in the installation plan
and that it is adequate to support the backup of the network server.
The backup system is generally a magnetic tape drive of one form or
another. The tape drive should be capable of backing up the con-
tents of the disk drives on the network server in a timely manner.
The capacity of the tape drive and the speed that data can be trans-
ferred to the tape drive are both of critical importance. If you deter-
mine that given the specifications of the tape drive, a full backup of
the network server would take 10 hours and you only have 4 hours
to perform the backup, for example, the tape drive is inadequate for
the job. You also can back up other devices, such as hard disk drives,
CD-R devices, and CD-RW devices. See Chapter 1 for a complete
discussion of backup systems.
The backup devices should be installed on a separate controller from
the disk drives of the network server, to provide peak performance
during the backup process.
SCSI Cables
Verify that the correct cables have been delivered to connect the
SCSI channel controller to the SCSI devices (disk drives). SCSI
cables differ distinctly from each other (for example, SCSI-1, SCSI-
2, and SCSI-3). The wide versions of SCSI-2 and SCSI-3 utilize dif-
ferent cables. Internal SCSI cables are generally ribbon cables, with
pin 1 identified by a colored (usually red) strip on the edge of the
ribbon cable. External SCSI cables are generally in a round bundle.
Verify that the SCSI cables have the correct number of connectors
for the network server configuration.
Make sure that the SCSI cables do not exceed the SCSI channel
length maximums. Also make sure that the SCSI cables have enough
connectors to allow all the SCSI devices to be attached. If you need
to attach four SCSI disk drives to a SCSI channel, for example, the
SCSI cable needs at least four connectors for the disk drives, plus
one where it can be attached to the SCSI adapter.
07 Serv+ Ch_02 7/10/01 9:48 AM Page 159
SCSI Adapter(s)
Verify that the correct SCSI adapter is available. Many network
servers have one or more SCSI adapters built in. Additional SCSI
adapters may be required to support the number of SCSI devices
that will used with the network server. Make sure that the SCSI
adapter and SCSI devices are of the same type of SCSI (SCSI-1,
SCSI-2, SCSI-3, and so forth). Make sure that the SCSI adapter’s
interface matches the bus that is in the network server for example,
EISA, PCI, or PCI-64.
RAID Controller(s)
If the network server is going to use the hardware version of RAID,
verify that the RAID controller is available. The RAID controller
should be delivered with configuration software. The RAID con-
troller must be configured before the network operating system can
be installed. Most RAID controllers are designed to support some
version of SCSI disk drives. Make sure that the RAID controller
matches the bus that is in the network server for example, EISA,
PCI, or PCI-64.
that all NICs are available. Make sure that the NIC supports the type
of network where the network server is going to be installed (Ethernet,
Token Ring, and so on). Multiple NICs might also be installed in a
single network server if it is to be attached to multiple networks.
Miscellaneous Hardware
Other hardware might be required for the network server. The net-
work server does require a video adapter to support the network
server monitor. There is no reason to spent lots of money getting a
fancy video adapter that has a large amount of video memory to
support extremely high video resolution and billions of colors. The
video monitor on a network server is generally utilized only to per-
form administrative functions on the network server. A video
adapter that can support VGA resolution of up to 1024 × 768 and
65,536 colors should be adequate for most network servers.
Many server hardware vendors have a specialized server management
adapter that can be purchased and installed in their network servers.
These devices generally monitor the health of the server hardware
and can be used by server management software running on a
designed management computer. Some of these devices have a built-
in modem, which will allow monitoring of the server hardware via a
dial-in connection.
A rack-mounted network server might include a keyboard/video/
mouse (KVM) switch to allow one keyboard, one video display, and
one mouse to be utilized by multiple network servers that are in a
single rack. The KVM switch allows the keyboard, mouse and video
display to switch (usually from the keyboard) among the network
servers in the rack. This saves space in the rack, because each network
server will not require its own keyboard, mouse, and monitor. Some
cost savings also is associated with sharing the keyboard, mouse, and
monitor, because you don’t have to buy them for each server.
A network server also should have some devices commonly found on
most desktop computer systems, such as a 3.5-inch floppy disk drive
and a CD-ROM drive or a DVD-ROM drive. These devices are
required to install operating system software, hardware drivers, and
other software onto the network server.
07 Serv+ Ch_02 7/10/01 9:48 AM Page 161
INSTALLING HARDWARE
. 1.2 Install hardware using ESD best practices (boards, drives,
memory, internal cables, etc.) These activities include:
• Mount the rack installation
• Cut and crimp network cabling
• Install Uninterruptible Power Supply
• Verify SCSI ID configuration and termination
• Install external devices (e.g., keyboards, monitors,
subsystems, modem rack, etc.)
• Verify power-on via power-on sequence
Once the server hardware specialist has verified all the hardware for the
network server, the next step is the actual installation of the hardware.
Electrostatic Discharge
Electrostatic discharge, better known as ESD, is a build up of elec-
tric charges on a body that can be transferred to another body upon
contact. This buildup of electric charge can easily damage the sensi-
tive components found inside a network server. While installing
components inside of a server chassis, it is very important that you
do not accidentally build up an electric charge that can be trans-
ferred to components inside the network server. You should follow
the list of ESD Best Practices included here anytime that you open
the chassis of the network server (to install, upgrade, or maintain
components in the chassis).
ESD can be caused by any of the following:
á People moving around (especially on carpet)
á Low humidity
á Unshielded cables
á Improper grounding
07 Serv+ Ch_02 7/10/01 9:48 AM Page 162
Know the ESD Best Practices While working on a network server, follow this list of the best prac-
You are likely to be asked about tices to ensure that you do not have a problem with ESD:
electrostatic discharge best prac-
tices on the exam. Know them. á Always use a static discharge device (static wrist strap, anti-
static mat) when you open the server chassis to access
components in the server.
á Always store components in static bags until you are ready to
use them.
á Before you touch anything inside a network server, touch the
housing of the power supply. This will discharge any static
charge that may have built up on your body.
á Use antistatic solutions on carpets in the server room.
á Use a room humidifier if the humidity is constantly below 50%.
á If possible, install antistatic carpet in the server room.
á Hold disk drives by the edges. Do not touch the circuit board
on the bottom of the disk drive.
Installing Processors
If you must install additional processors in a network server, be sure
that you follow the ESD best practices. Always follow the server hard-
ware vendor’s installation instructions. You must check several factors
about additional processors before installing the additional processors.
TIP
A stepping is the Intel term used to describe different versions of a Stepping Know that multiple
given processor. Intel recommends that processors in a multi- processors in a network server
EXAM
processor server should have steppings that differ by no more than must meet the (N+1) stepping
requirements.
one. Intel refers to this as (N+1) stepping. It is best that the processors
have the same identical stepping number.
For more information about steppings, see Intel’s Web site at
www.intel.com/newsgroups/mips/mip_2.htm#five.
Buses
The server hardware specialist does not install the buses in the net-
work server, but needs to be aware of the buses that are in the net-
work server to install additional adapters in the network server. Most
servers have one or more PCI or PCI-64 buses. The multiple buses
allow the server to transport more data between the NICs and the
network server or the SCSI controller and the network server. The
PCI bus is 32–bits wide, which means that it can transfer 4 charac-
ters or bytes of data per clock cycle. (32 bits divided by 8 bits per
byte equals 4 bytes.) The PCI-64 bus is 64-bits wide, which means
that it can transfer 8 characters of data per clock cycle. The PCI and
PCI-64 buses operate at 33 Megahertz. This means that a PCI bus
has a maximum theoretical transfer rate of 32MBps (33 million
hertz × 4 bytes/second). A PCI-64 bus has a maximum theoretical
transfer rate of 264MBps (33 million hertz × 8 bytes/second). Some
newer PCI buses operate at 66 Megahertz, thus doubling the trans-
fer rate. The trade-off for this faster bus speed is usually fewer con-
nections available on the bus. PCI buses that operate at 100MHz
will soon be available.
Peer Buses
A network server with peer buses has multiple buses that act indepen-
dently. Most peer bus systems are dual peer (two buses) or triple peer
(three buses). This means that each bus has the full transfer rate avail-
able. On a server that will be transferring large amounts of data to and
from the disk drives, as well as transferring large amounts of data to
and from the network, can have this workload balanced between the
peer buses. You just place the disk controller on one bus and the NIC
on the other bus. Figure 2.12 shows a diagram of a peer bus.
07 Serv+ Ch_02 7/10/01 9:48 AM Page 164
Bridge Bridge
A dual-peer PCI bus provides two separate buses that can each
operate at full speed, effectively doubling data transfer capacity.
To take advantage of the two buses, install adapters that require
high data-transfer capability on separate buses. (For instance, put
a 100BASE-T network adapter on one bus and put a SCSI con-
troller on the other bus.)
Bridged Buses
EXAM TIP
Buses Know the difference A server with bridged buses has multiple buses that act as one. This
between a peer bus and a bridged is done to increase the number of PCI slots available on a server.
bus. Both buses share the available transfer speed. Placing the disk con-
troller on one bus and the NIC on the other bus has no load balanc-
ing benefit because the buses are essentially one. Figure 2.13 shows a
diagram of a bridged PCI bus.
Memory Installation
If you must install additional memory into the network server, be
sure that you follow the ESD best practices listed earlier in this
chapter. Always follow the server hardware vendor’s installation
07 Serv+ Ch_02 7/10/01 9:48 AM Page 165
System Bus
FIGURE 2.13
Bridged PCI bus.
Processor Memory
Bridge
Bridge
IDE/ATA Disks
IDE disk drives and controllers (also known as ATA disk drives and
controllers) are limited to two channels with a maximum of 2 disk
drives per channel. The IDE/ATA channel is a 40-conductor ribbon
cable that is a maximum of 18 inches long. The short channel
length limits IDE/ATA to devices that are internal to the network
server. That is a total of 4 disk drives per IDE/ATA disk controller.
These disk controllers rely on the processor of the computer to per-
form most of the disk input/output operations.
Each disk drive that is attached to an IDE/ATA disk controller must
have a jumper set on the back of the disk drive to specify the role the
disk will perform in relationship to the other disk on the same chan-
nel. A single IDE disk is usually set to the master role. If a second
IDE disk is attached to the same channel as an existing IDE disk
drive that is set to Master, it must be set to the slave role. In essence,
the disk hardware on the Master disk drive controls both the Master
and Slave disk drives. A third option called Cable Select (CSEL) will
allow the IDE adapter to select which IDE disk drive will function as
Master and which IDE disk drive will function as Slave. If both IDE
disk drives that are on a single channel are set to Master, or if both
are set to Slave, the disk subsystem will not work. No industry stan-
dard applies to the jumper settings for IDE disk drives, so read your
drive documentation carefully. Figure 2.14 shows a diagram of the
selections available for a typical IDE disk drive.
07 Serv+ Ch_02 7/10/01 9:48 AM Page 167
FIGURE 2.14
Representative IDE disk drive jumper settings.
Drive 1:
Master
Drive 2:
Slave
EIDE/ATA-2 Disks
NOTE
drive will function as Master and which EIDE/ATA-2 disk drive will
function as Slave. If both EIDE/ATA-2 disk drives that are on a sin-
gle channel are both set to Master, or are both set to Slave, the disk
subsystem will not work.
SCSI Disks
The majority of server hardware will use Small Computer Systems
Interface (SCSI) disk drives for a multitude of reasons, including
the following:
07 Serv+ Ch_02 7/10/01 9:48 AM Page 169
E X A MT I P
SCSI buses can handle a total of 8 devices per channel, which are Terminating SCSI Know that
numbered from 0 through 7. (Some versions of SCSI support a total both ends of a SCSI bus must
of 16 devices per channel, which are numbered 0 through 15.) Each be terminated.
device on a SCSI channel must have a unique SCSI ID. SCSI ID
numbers are generally set by jumpers on the SCSI disk drive itself.
Duplication of the SCSI IDs on a SCSI channel can cause the chan-
nel to be inaccessible. If you want to be able to boot from a SCSI
disk drive, give it a SCSI ID of 0 or 1.
Other rules for SCSI are pretty simple. Each end of the SCSI bus
must be terminated. Generally the SCSI controller itself is on one
end of the SCSI bus and it usually has onboard termination. The
other end of the SCSI cable is either terminated by a resistor on the
last disk drive on the chain or a physical terminator on the end of
the SCSI bus. Figure 2.15 shows an example of SCSI termination. SCSI
Controller
There are three types of SCSI termination: passive termination,
active termination, and forced perfect termination (FPT). Passive
termination is cheap to implement, but should be used only for
lower speed SCSI channels with relative short cables. Active termina-
tion is preferred and can handle much higher speeds and longer
cables. Forced perfect termination is the best method and can be Terminated Not Terminated
Terminated
used for even the highest speed SCSI implementation. It also is the
most complex to implement and therefore the most costly. However, FIGURE 2.15
the extra cost will result in a much more reliable SCSI system. SCSI bus termination.
07 Serv+ Ch_02 7/10/01 9:48 AM Page 170
SCSI-1
Originally just known as SCSI, it was used by many Apple computers
in the early 1980s. By today’s standards it was rather slow. The SCSI
bus ran at 5MHz using an 8-bit data path. This allowed a data
transfer rate of 5MB per second.
07 Serv+ Ch_02 7/10/01 9:48 AM Page 171
WA R N I N G
The SCSI-2 internal cable was a ribbon cable that was attached to Differential Danger Attaching a
the disk controller by a 50-pin connector. Many early SCSI con- device using the differential (HVD)
trollers used a DB-25, 25-pin connector for external SCSI devices. signaling method can seriously
The termination for the SCSI-1 was usually a set of 3 resistors on damage devices designed to use
the single-ended or low-voltage dif-
the SCSI controller (assuming it was at the end of the SCSI bus) or
ferential signaling systems.
a set of three resistors on the last SCSI disk drive on the bus, or an
actual terminator attached to the end of the SCSI bus. Maximum
cable length of SCSI-1 is 6 meters.
SCSI-2
SCSI-2 uses two different signaling systems, known as single-ended
interface and differential interface. The two signaling systems are
incompatible and cannot be mixed on the same SCSI bus. Make
sure that all devices, including the SCSI-2 controller, are all using
either single-ended interface or all using the differential interface.
Due to bus length restrictions, single-ended SCSI-2 cabling is usu-
ally found inside a server chassis. The differential interface allows for
longer cable lengths and is generally found connecting the server to
an external SCSI device. SCSI-2 uses the same 50-pin connector on
the internal SCSI cable that is used by SCSI-1 devices.
However, SCSI-2 also has a variant called Wide SCSI-2, which can
transfer 16 bits at a time as opposed to the 8 bits at a time used by
normal SCSI-1 and normal SCSI-2. This extra bus width requires
the use of a 68-pin connector. Wide SCSI-2 allows for 16 devices on
the SCSI-2 channel, whereas normal SCSI-2 (also called narrow
SCSI-2) and SCSI-1 only allow 8 devices on the SCSI channel.
identify 50-pin, 68-pin, and 80-pin tor that combines the functions of a normal SCSI connector, a
SCSI connectors.
power connector, and the SCSI ID’s jumpers into one connector.
EXAM
FIGURE 2.17 25 1
50-pin SCSI connector.
50 26
FIGURE 2.18 34 1
68-pin SCSI connector.
68 35
Pin 80
07 Serv+ Ch_02 7/10/01 9:48 AM Page 173
SCSI-3
SCSI-3 is the latest standard of the SCSI family. It combines all the
best features of the previous SCSI standards. It uses LVD, which
uses differential signaling, and supports up to 15 devices on a single
cable, which can be up to 12 meters long. SCSI-3 supports three
different bus speeds, known as Ultra (20MHz), Ultra2 (40MHz),
and Ultra3 (double-clocked 40MHz). There are both narrow (8-bit)
and wide (16-bit) implementations of the three SCSI-3 bus speeds.
Ultra SCSI-3 and Ultra2 SCSI-3 both use 50-pin connectors. The
wide variants (Wide Ultra SCSI-3 and Wide Ultra2 SCSI-3) as well
as Ultra3 (also known as Ultra160 SCSI-3) use 68-pin connectors.
All versions of SCSI-3 require active termination. Table 2.1 summa-
rizes the maximum cable lengths for various types of SCSI.
TABLE 2.1
M A X I M U M SCSI B U S L E N G T H BY TYPE
Type of SCSI Length
Standard 6 meters
Fast 3 meters
Wide-Ultra 1.5 meters
Low-voltage differential 12 meters
Differential 25 meters
Installing SCSI
Step-by-Step 2.2 walks you through the process of installing SCSI
components.
STEP BY STEP
2.2 Installing SCSI Components
1. Set the SCSI IDs on the disk drives (see Table 2.2).
2. Insert SCSI controller into the server.
3. Install SCSI disk drives into the server.
continues
07 Serv+ Ch_02 7/10/01 9:48 AM Page 174
continued
4. Connect the SCSI ribbon cable to SCSI controller. (Red
strip on the ribbon cable should be attached to pin 1 on
the SCSI controller.)
5. Connect the SCSI ribbon cable to SCSI disk drives. (Red
strip on the ribbon cable should be attached to pin 1 on
the SCSI connector on the disk drive.)
6. Terminate both ends of the SCSI bus.
7. Connect the power connector to the SCSI disk drives.
8. Document all settings.
Table 2.2 shows the various jumper settings for setting the SCSI ID
on SCSI devices.
TABLE 2.2
SCSI A D D R E S S J U M P E R S E T T I N G S
SCSI ID Jumper 4 Jumper 3 Jumper 2 Jumper 1
0 OFF OFF OFF OFF
1 OFF OFF OFF ON
2 OFF OFF ON OFF
3 OFF OFF ON ON
4 OFF ON OFF OFF
5 OFF ON OFF ON
6 OFF ON ON OFF
7 OFF ON ON ON
8 ON OFF OFF OFF
9 ON OFF OFF ON
10 ON OFF ON OFF
11 ON OFF ON ON
12 ON ON OFF OFF
13 ON ON OFF ON
14 ON ON ON OFF
15 ON ON ON ON
07 Serv+ Ch_02 7/10/01 9:48 AM Page 175
Only Wide SCSI devices have 4 jumpers. All other SCSI devices
have 3 jumpers. Notice that if you substitute the numeral 1 for all
the ON settings and the numeral 0 for all the OFF settings, you are
counting from 0 to 15 in binary.
Ethernet
The vast majority of networks today are based on a version of
Ethernet. The two most common types of Ethernet are known as
10BASE-T and 100BASE-T. 10BASE-T Ethernet operates at
10Mbps. 100BASE-T Ethernet operates at 100Mbps and is known
as Fast Ethernet. 1000BASE-T is becoming available and operates at
1Gbps (1000 megabits per second) and is known as Gigabit
Ethernet. Luckily 10BASE-T and 100BASE-T use the same type of
network cabling, unshielded twisted-pair (UTP). The connector on
the end of the UTP is known as an RJ-45 connector. The twisted-
pair cable contains 4 pairs of 2 wires each that are twisted around 8 1
each other. One end of the UTP cable plugs into an RJ-45 jack on
the back of the NIC in the server. The other end of the UTP cable
plugs into a network device (either a hub or a switch). The maxi-
mum distance from the hub or switch to the NIC is 100 meters. FIGURE 2.20
Figure 2.20 shows an example of an Ethernet connection. Ethernet RJ-45 connector.
07 Serv+ Ch_02 7/10/01 9:49 AM Page 176
Token Ring
Token-Ring networks can use either the same type of cable that is
used by Ethernet network (UTP) or a more expensive type of cable
known as shielded twisted-pair (STP). If UTP cable is used, the
cable plugs into an RJ-45 jack on the back of the NIC in the server.
The other end of the cable is plugged into an RJ-45 connector on a
device known as a multi-station access unit (MSAU). If STP cable is
used, the cable plugs into a DB-9 connector on the back of the NIC
(only 4 pins are used). The other end of the STP cable has an IBM
Universal Data Connector on it that plugs into an MSAU. The dis-
tance from the NIC to the MSAU in either case is limited to 100
meters. Token Ring operates at either 4Mbps or 16Mbps. Figure
2.21 shows an example of a Token-Ring connection.
Network Cabling
Most networks today are built using UTP cable. The connector on
the end of the cable is known as an RJ-45 connector. UTP cable has
a total of 8 wires in it (4 pairs of 2 wires each). The RJ-45 connector
terminates all 8 wires in a clear plastic connector. The 8 wires are
Connectors for Token-Ring network cable. [Connects to MultiStation Access Unit (MSAU)]
07 Serv+ Ch_02 7/10/01 9:49 AM Page 177
color coded to allow them to be placed in the correct order when the
UTP cable is terminated. A crimping tool is used to terminate a UTP
cable in an RJ-45 connector. Figure 2.22 shows a crimping tool.
NIC to Hub
The cable used to connect an Ethernet NIC to an Ethernet hub (or FIGURE 2.22
A UTP crimping tool.
switch) is called a straight-through cable. The 8 wires are in the same
order (1–8) at each end of the cable. This is the most common type
of Ethernet cable. It also is sometimes called a patch cable. Figure
NOTE
2.23 shows an example of wiring a straight-through UTP cable. Add to Tool Kit A server hardware
specialist’s tool kit should include a
supply of RJ-45 connectors and a
Hub to Hub crimping tool.
If you need to connect two Ethernet hubs together in a network, you
have two choices. If the Ethernet hub has a special RJ-45 jack known
as an uplink port (also sometimes labeled MDI-X ), you can use a
EXAMTIP
standard Ethernet straight-through UTP cable. This is just like the
cable normally used to attach an Ethernet NIC to an Ethernet hub. Ethernet Wiring Know the wiring
for an Ethernet UTP straight-through
The other choice comes into play when the Ethernet hubs do not cable.
have an uplink port. You can connect two Ethernet hubs together by
using a special UTP cable known as a crossover-cable. In a cross-over
cable, the 8 wires in the UTP cable are wired differently. In a cross-
over cable, wire 1 goes to wire 3 on the other end, wire 2 goes to NOTE Add to Tool Kit A server hardware
wire 6 on the other end. Wires 4, 5, 7, and 8 go straight through. specialist’s tool kit should include a
Figure 2.24 shows an example of a cross-over cable. 10-foot straight-through cable.
FIGURE 2.24
Ethernet RJ-45 cross-over cable. Pins 4, 5, Transmit (1&2)
7 and 8 Receive (3&6)
are not used
Pins 4, 5,
Receive (3&6) 7 and 8
Transmit (1&2) are not used
UPS Cabling
Monitor
The monitor is connected to the video adapter via a 15-pin HD-15
connector. If the CRT monitor is to be rack mounted, it usually sits
on a specially designed shelf in the rack (see Figure 2.26).
Keyboard
The keyboard plugs into the keyboard port on the server. There is
no difference between plugging a keyboard into a server and plug-
ging a keyboard into a standard desktop computer. Care must be
taken not to reverse the mouse and keyboard connections, because
they use the same type connector. If the keyboard is a special rack-
mount keyboard, it will fit into a rack-mounted drawer in the rack.
The rack-mounted keyboard may have an integrated trackball or
touch pad to substitute for a mouse (see Figure 2.27).
Mouse
The mouse plugs into the mouse port (sometimes called a PS/2
mouse port) on the server. There is no difference between plugging a
mouse into a server and plugging a mouse into a standard desktop
computer. Care must be taken to not reverse the keyboard and
mouse connections, because they use the same type PS/2 connector.
A mouse with a Universal Serial Bus (USB) connector might also be
utilized on a network server.
07 Serv+ Ch_02 7/10/01 9:49 AM Page 180
FIGURE 2.26
CRT monitor on shelf in rack.
FIGURE 2.27
Rack-mounted keyboard.
07 Serv+ Ch_02 7/10/01 9:49 AM Page 181
Video Adapter
The video adapter for the server may be either built in or an adapter
that plugs into an available bus slot. No need for a fancy video
adapter with lots of video memory. The server video display is usu-
ally used only for simple administrative tasks that do not require
high resolution.
CD-ROM/DVD-ROM Drive
Most server hardware includes either a CD-ROM or DVD-ROM
drive. This is usually required for operating system and other soft-
ware installation.
Power Supply
Server hardware may have multiple hot-swappable power supplies.
Having two or even three power supplies allows the power supplies
to balance the electrical load. If one power supply fails, the other
power supply can and will handle the entire electrical load. The
failed power supply can be replaced with a new power supply with-
out taking the network server offline.
07 Serv+ Ch_02 7/10/01 9:49 AM Page 182
Cooling Fan
Keeping the inside of the server chassis relatively cool is of major
importance. If components inside the server chassis get too hot, they
will tend to fail. Server hardware has cooling fans to keep air circulating
around the components. If you have only one cooling fan and it fails,
the inside of the server chassis will overheat and the server will fail.
Many hardware manufacturers provide the ability to add a redundant
cooling fan, which is hot swappable in case of a cooling fan failure.
Redundant NIC
Having a redundant NIC in the network server will allow the net-
work server to keep communicating over the network even if a NIC
fails. This usually is accomplished by having two identical NICs in
the server, only one of which is actually communicating over the net-
work. A special software driver for the redundant NICs constantly
monitors the network communications and if the software driver
determines that the NIC currently being used to communicate over
the network has failed, it will automatically switch to the other NIC.
Server-Rack Installation
If the network server you are installing is a rack-mount server, you
must take a few special steps. You must follow the rack-installation
instructions provided by the server vendor. Not only do you have to
have all rack-mount components (server, UPS, monitor, keyboard/
mouse, and so on), but there are also recommendations that you
should follow as to the order that equipment is installed in the rack.
Also, each rack-mount component has special supports that must be
installed on the component and the rack to later provide access to
the components installed in the rack. Because servers installed in the
rack must be slid out of the rack for you to gain access to the inside
of the chassis, special cable management arms are attached to the
back of the components. This allows the server chassis, for example,
to be slid out of the rack for access without having to disconnect any
of the cables attached to the rear of the server (for instance, network
cable, UPS monitor cable, SCSI cables). Figure 2.28 shows a picture
of a cable management arm.
07 Serv+ Ch_02 7/10/01 9:49 AM Page 183
If more than one network server is installed into a single server rack,
a keyboard/video/mouse (KVM) switch should also be installed to
allow the use of one keyboard, one video monitor, and one mouse to
control all network servers in a single rack. Figure 2.29 shows the
wiring for a KVM switch.
FIGURE 2.28
Cable management arm in server rack.
MON
ITOR
VGA
1
VGA
2
VGA
3
Keyboard 1 VGA
Mouse 1
4
Keyboard 2
Mouse 2
Keyboard 3
Mouse 3
Keyboard 4
Mouse 4
FIGURE 2.29
Wiring a KVM switch.
07 Serv+ Ch_02 7/10/01 9:49 AM Page 184
Power-On Test
NOTE
524288 KB Detected
COMPAQ System BIOS - E16 (12/30/1996)
Copyright 1982, 1996 Compaq Computer Corporation. All rights reserved.
FIGURE 2.30
SCSI adapter finds SCSI devices during
power up.
07 Serv+ Ch_02 7/10/01 9:49 AM Page 185
CHAPTER SUMMARY
The chapter discussed verification of the installation plan and then
KEY TERMS
focused on the actual installation of the hardware for a network server.
• ANSI
The discussion began with the verification of the installation plan.
• ATA You read about the need to check the installation site for power,
• ATA-2 cooling, and network connectivity. You also learned to check the
network server components that are available and compare them to
• Bus the network server components as specified in the installation plan.
• Cross-over cable The discussion then turned to the process of installing the compo-
• Differential nents into the network server chassis. These components include
processors, memory, adapter cards (such as SCSI, RAID, FC-AL,
• EIDE
and network). After installation of the components, you learned to
• Electrostatic discharge run a preliminary test of the server by performing a power on.
• Ethernet In the next chapter, you will configure the components and install
the network operating system.
07 Serv+ Ch_02 7/10/01 9:49 AM Page 187
CHAPTER SUMMARY
• IDE
• L2 cache
• Low-voltage differential
• Keyboard/video/mouse switch
• PCI bus
• Rack
• RAID
• RJ-45
• SCSI
• SCSI-1
• SCSI-2
• SCSI-3
• Single ended
• Standard rack unit
• Stepping
• STP
• Straight-through cable
• Token Ring
• Ultra ATA
• Ultra DMA
• Ultra SCSI
• Ultra2 SCSI
• Uninterruptible power supply
• UTP
• Wide Ultra SCSI
07 Serv+ Ch_02 7/10/01 9:49 AM Page 188
A P P LY Y O U R K N O W L E D G E
Exercises 3. Does the disk drive use a 50-pin, a 68-pin, or an
80-pin connector?
2.1 Becoming Familiar with IDE Disk Drives 4. Attach the cable to the disk drive.
You are probably already familiar with IDE drives and
their installation in a desktop PC. Just in case you are
not, this exercise runs you through some drive basics.
Review Questions
And even if you are familiar with IDE drives, it doesn’t 1. What are the different types of server chassis?
hurt to refresh your memory.
2. How big is one rack unit?
Estimated Time: 10 minutes.
3. How do you convert watts to volt-amps?
1. Get an IDE/ATA or EIDE/ATA-2 disk drive
4. What is the rule on processor steppings when
and cable.
multiple processors are installed in a server?
2. Examine the disk drive. You should notice a
5. What is electrostatic discharge?
sticker on the back of the disk drive that gives you
instructions on how to set the disk drive to be a 6. When do you use a UTP cross-over cable?
single drive, to be a Master drive, or to be a Slave 7. How many terminators are on a properly termi-
drive. You might also see whether the disk drive nated SCSI bus?
supports CSEL. The jumper settings and their
names are not standardized and therefore differ 8. How many ATA-2 disk drives can be connected
from one manufacturer to another. to one ATA-2 channel?
3. Attach the cable to the disk drive. 9. How many Wide SCSI-2 disk drives can be
connected to one Wide SCSI-2 channel?
A P P LY Y O U R K N O W L E D G E
Exam Questions A. 10 meters
A. 3 meters C. Rack-mount
A. 32KB D. Memory
A P P LY Y O U R K N O W L E D G E
10. How many wires are in a UTP cable that is B. 36 inches
terminated with an RJ-45 connector?
C. 72 inches
A. 2
D. 144 inches
B. 4
15. If the first disk drive on an EIDE/ATA-2 channel
C. 6 is set to the master role, what role must a second
disk drive take when attached to the same channel?
D. 8
A. Master
11. What is the name of the tool that is used to
terminate a UTP cable in an RJ-45 connector? B. Slave
A. Terminator C. CSEL
B. Pliers
C. Crimper Answers to Review Questions
D. Pincher 1. There are three major types of chassises for
servers: tower, wide tower, and rack mount. For
12. How should the jumpers on a SCSI disk drive be
more information, see the section “Server
set in order for it to have a SCSI ID of 6?
Chassis.”
A. ON ON ON
2. One standard rack unit is 1.75 inches. For more
B. ON ON OFF information, see the section “Server Rack.”
C. ON OFF OFF 3. A good estimate to convert watts to volt-amps is
to multiply watts by 1.5. For more information,
D. OFF OFF OFF
see the section “UPS Sizing Worksheet.”
13. Which of the following SCSI connectors com-
4. When multiple processors are installed in a net-
bines the normal SCSI control lines, SCSI ID
work server the steppings (versions) must be
setting, and power into a single connector?
within 1 number of each other (known as (N+1)
A. 50-pin connector stepping). For more information, see the section
B. 68-pin connector “Match Processor Stepping.”
A P P LY Y O U R K N O W L E D G E
7. A SCSI bus must be terminated on both ends, 15. An uninterruptible power supply should be large
therefore two terminators are required. For more enough to keep the network server equipment
information, see the section “SCSI Disks.” running long enough to allow a graceful shut-
down on the network server. For more informa-
8. Two ATA-2 disk drives can be connected to a sin-
tion, see the section “Verifying the
gle ATA-2 channel. One disk assumes the master
Uninterruptible Power-Supply Size.”
role and the other disk assumes the slave role.
For more information, see the section
“EIDE/ATA-2 Disks.” Answers to Exam Questions
9. A wide SCSI channel can support 16 devices; one 1. D. SCSI uses 50-pin, 68-pin, and 80-pin (SCA)
must be the SCSI controller, the other 15 devices connectors. For more information, see the section
can be disk drives. For more information, see the “SCSI Disk Types.”
section “SCSI-2.”
2. D. Differential SCSI has a maximum cable
10. A KVM switch (keyboard/video/mouse) permits length of 25 meters. For more information, see
control of several servers from a single keyboard, Table 2.1.
video monitor, and mouse. The alternative is to
use a keyboard, video monitor, and mouse for 3. C. Typical processor cache sizes are 256KB,
each server. For more information, see the section 512KB, 1024KB, and 2048KB. For more
“Miscellaneous Hardware.” information, see the section “Match Processor
Cache Size.”
11. The three types of SCSI bus termination are pas-
sive, active, and forced perfect. See “SCSI Disks.” 4. C. Seven devices can be attached to a narrow
SCSI channel. (There are actually 8 devices, but
12. The maximum length of a low-voltage differen- one of them is the SCSI controller itself.) For
tial (LVD) SCSI cable is 12 meters. For more more information, see the section “SCSI Disks.”
information, see Table 2.1.
5. B. The maximum cable length between an
13. Redundant power supplies can balance the Ethernet hub and an Ethernet NIC in a com-
electrical load and can power the network puter is 100 meters. For more information, see
server in the event of a single power-supply the section “Ethernet.”
failure. For more information, see the section
“Power Supply.” 6. C. A rack-mount chassis is designed to use a
minimum amount of floor space. In normal prac-
14. A hardware compatibility list (HCL) lists the tice, several network servers can be installed into
hardware that has been verified as working with a a single rack. For more information, see the
particular network operating system. For more section “Server Rack.”
information, see the section “Verifying Hardware
Compatibility with the Operating System.”
07 Serv+ Ch_02 7/10/01 9:49 AM Page 192
A P P LY Y O U R K N O W L E D G E
7. C. The video adapter is the least important of the 12. B. The SCSI jumpers set to ON ON OFF repre-
four hardware components listed for a network sents the SCSI ID 6. For more information, see
server. For more information, see the section Table 2.2.
“Video Adapter.”
13. C. The 80-pin SCA SCSI connector combines
8. C. SCSI supports both internal devices and exter- the SCSI control signals, the SCSI ID settings,
nal devices. For more information, see the section and power into a single connector. For more
“SCSI Disks.” information, see the section “SCSI-2.”
9. A. A magnetic tape drive is the most common 14. A. The maximum IDE/ATA/EIDE/ATA-2 cable
device used for backups. For more information, length is 18 inches. For more information, see the
see the section “Backup System.” section “IDE/ATA Disks.”
10. D. There are 8 wires (4 pairs) in a UTP cable that 15. B. The second EIDE/ATA-2 disk drive must
is terminated with an RJ-45 connector. For more assume the slave role if the first disk drive is set
information, see the section “Network Cabling.” to the master role. For more information, see the
section “EIDE/ATA-2 Disks.”
11. C. The tool that is used to terminate a UTP
cable in an RJ-45 connector is called a crimper.
For more information, see the section “Network
Cabling.”
A P P LY Y O U R K N O W L E D G E
7. IDE/EIDE Disk Drives FAQ 12. Electronic Dictionaries on the World Wide
Web
www.faqs.org/faqs/pc-hardware-
faq/enhanced-IDE/part1 Whatis.com
www.faqs.org/faqs/pc-hardware- www.whatis.com
faq/enhanced-IDE/part2/
Webopedia.com
8. Disk Drive Specifications and Jumper www.webopedia.com
Settings
13. Free On-Line Dictionary Of Computing
www.thetechpage.com/cgi-bin/default.cgi
(FOLDOC)
General Computer Hardware Sites
http://foldoc.doc.ic.ac.uk/foldoc/
index.html
9. The PC Guide
http://www.pcguide.com
07 Serv+ Ch_02 7/10/01 9:49 AM Page 194
08 Serv+ Ch_03 7/11/01 10:01 AM Page 195
OBJECTIVES
3
Although a server hardware specialist would not
actually configure the network, you need to under-
stand network configuration well enough to verify
that the server is connecting to the network. C H A P T E R
OBJECTIVES OUTLINE
OUTLINE S T U DY S T R AT E G I E S
INTRODUCTION
This job dimension covers the configuration of network server hard-
ware and software and it is the topic of approximately 18% of the
Server+ Certification exam.
After installation of the hardware components for the network
server, the next step is to configure the hardware components as
specified in the installation plan. This includes checking and if
necessary upgrading the BIOS of major components, system, SCSI
controller, and RAID controller. If hardware-based RAID is being
implemented on the network server, it must be configured (select
RAID levels and disk partitions) before any software can be installed
on the network server. The network operating system is installed
next and any “patches” or “updates” to the network operating system
are applied. Drivers specific to the manufacturer hardware in the
network server must be installed to support the manufacturer-
specific hardware. External devices, such as a UPS, must be config-
ured. Network server management software—such as SNMP,
backup, system monitoring agents, and antivirus software—must
be installed and configured. After all software has been installed
and configured, a performance baseline measurement should be
taken and the complete hardware and software configuration should
be documented.
The system BIOS is the most well known of the several BIOSs usually
found in the network server. On occasion the network server hardware
manufacturer might update the system BIOS. These changes or
updates to the BIOS are usually done to add features to the system
BIOS (such as Plug and Play support, or support for large hard disk
drives), correct BIOS bugs (such as the recent Year 2000 date prob-
lem), or support new hardware. Most system BIOSs can be upgraded
by a program supplied by the server hardware manufacturer.
Small Computer System Interface (SCSI) controllers have a BIOS
on the controller itself. Although not updated as frequently as a
system BIOS, on occasion the SCSI controller manufacturer might
upgrade the SCSI controller BIOS to correct bugs in the original
BIOS or to add new features.
RAID controllers also contain a BIOS. Again, the BIOS on a
RAID controller will have updates issued by the RAID controller
manufacturer.
WA R N I N G
ber of the system BIOS usually displays. Check the network server Take Care When Flashing That
vendor’s web site to determine whether the version of the system BIOS Loss of power to the net-
BIOS installed on your network server is the latest system BIOS work server while its flash BIOS is
available for the model of the network server that you have. If a being upgraded could result in a
totally unusable server, requiring
newer version of the system BIOS is available on the vendor’s web
replacement of the system board.
site, download the upgrade and follow the vendor’s instructions to
This is a good example of when you
update the system BIOS on the network server. Most network would want to make sure that the
servers have a BIOS that is flashable, meaning that it can be easily network server is plugged into a
erased and updated via software. functioning uninterruptible power
supply. This will prevent the flash
upgrade of the BIOS from becoming
Checking the SCSI Controller BIOS a total disaster if an electrical
power outage occurs during the
During the network server’s power up, the SCSI controller displays
upgrade.
the current version of the BIOS on the SCSI controller. It is advis-
able to check the vendor’s web site for the latest BIOS available for
the model of SCSI controller that you have. If the SCSI controller
has a flashable BIOS, it can be updated via software. Some SCSI
controllers require that a new chip containing the new BIOS be pur-
chased to replace the BIOS chip on the SCSI controller.
08 Serv+ Ch_03 7/10/01 9:50 AM Page 200
server. Because disk drives are mechanical devices, they will fail at
some time. It is not a matter of if they will fail, it is just a matter of
when they will fail. The term hardware-based RAID means that the
disk drives in the network server have RAID implemented by a spe-
cial disk controller, the RAID controller. Some network operating
systems can implement software-based RAID at the expense of addi-
tional load on the network server processor. Most RAID controllers
are designed to use SCSI disk drives. However, at least one disk con-
troller manufacturer makes a RAID controller that uses EIDE/ATA-2
disk drives. For an in-depth discussion of the different RAID levels,
see Chapter 1, “0.0—Planning and Specifications,” and Chapter 8,
“7.0—Disaster Recovery.”
The RAID disk controller has its own processor to implement the
RAID configuration, thus relieving the network server processor of
this task.
The configuration of the RAID controller in the network server is
accomplished by software provided by the network server (or RAID
controller) vendor. Although vendor specific, all the software works
basically the same way. It enables you to see the disk drives attached
to the RAID controller. You select the disk drives that you want to
utilize and specify the version of RAID that you want to implement
using the selected disk drives. The software then “prepares” the disk
drives to implement the RAID solution. For example, you might
pick two physical disk drives and tell the RAID configuration soft-
ware to use these two disk drives to implement a RAID 1 (mirror-
ing) solution. The RAID controller would tell the network server
operating system that there is a single logical disk drive. (When in
actuality, the RAID controller is reading and writing to two physical
disk drives.)
In another example, you might select five physical disk drives and
tell the RAID configuration software to use these five physical disk
drives to implement RAID 5 (disk striping with parity). The RAID
controller would tell the network server operating system that there
is a single logical disk drive. (When in actuality, the RAID controller
is reading and writing data in blocks across all five disk drives in the
RAID 5 disk array.)
In yet another example, you might select the same five physical disk
drives and tell the RAID configuration software to use these five
physical disk drives to implement RAID 5 (disk striping with parity).
08 Serv+ Ch_03 7/10/01 9:50 AM Page 202
Further, the single logical disk drive could be “partitioned” into two
NOTE
What Is a Partition? A disk partition partitions by the RAID configuration software. The network operat-
is a portion of a physical disk drive (or ing system would see two logical disk drives (one on each partition),
in the case of a RAID system, a por- whereas in actuality the RAID controller is reading and writing data
tion of a logical disk drive) that can be
in blocks across all five disk drives in the RAID 5 disk array.
identified (by drive letter or volume
name) and used separately from the Figure 3.1 shows an example of how a RAID controller manages
other portions of the physical disk disk drives and shows the logical disk drive to the network server
drive. operating system.
Logical Disk
Drive
FIGURE 3.1
RAID controller disk drive presentation.
08 Serv+ Ch_03 7/10/01 9:50 AM Page 203
TIP
RAID 1 has two different implementations: disk mirroring and disk
duplexing. In disk mirroring, everything written to one disk is also RAID 0 Is Not Fault Tolerant
Know that RAID 0 is not fault toler-
EXAM
written to a second disk. Reading can be performed from either
ant. It is used only to improve disk
disk. Because data is duplicated on two different disk drives, the fail-
input/output performance.
ure of one disk drive does not represent a catastrophe, but merely an
inconvenience. The network server will keep running using the sin-
gle working disk drive. When time permits, the broken disk drive
can be replaced and the data on the working disk drive copied to the Drive 1 Drive 2
Disk Controller
FIGURE 3.3
RAID 1 (disk mirroring).
08 Serv+ Ch_03 7/10/01 9:50 AM Page 204
IN THE FIELD
NOTE
RAID 5 Configuration
Block 1 Block 2 Block 1 Block 2
RAID 5 uses a much more complicated scheme to provide fault tol-
Block 3 Block 4 Block 3 Block 4 erance in the case of a single disk failure. Chapter 1 provides an in-
Block 5 Block 6 Block 5 Block 6
depth discussion of RAID 5.
RAID 5 requires a minimum of three disk drives to implement. The
disk drives that comprise a RAID 5 solution are often referred to as a
RAID 5 array. The failure of a single disk drive does not cause the
Disk Controller Disk Controller
network server to fail. The “missing” information that was on the
failed disk can be re-created on-the-fly using the information on the
remaining disks. The failed disk drive should be replaced as quickly
as possible. RAID 5 cannot survive the failure of a second disk drive
FIGURE 3.4 after one disk drive has failed. Because of this fact, some RAID
RAID 1 (disk duplexing).
systems allow for the configuration of a “hot spare” disk drive in the
08 Serv+ Ch_03 7/10/01 9:50 AM Page 205
RAID system. A hot spare disk drive is powered up and running, but Disk 1 Disk 2 Disk 3
contains no data; it is just waiting for a drive in the disk array to fail
so that it can be used. When the failure of a disk drive in the RAID Block 1 Block 2 Block 3
array occurs, the RAID system starts rebuilding the data that was on
the failed drive on the hot spare disk drive. This hot spare methodol- Parity 1 Block 4 Block 5
ogy minimizes the amount of time it takes to get the RAID rebuilt Block 6 Parity 2
EXAM TIP
drives in the RAID 5 array. RAID 1 Disk Overhead Is 50%
Know that the disk overhead for
The total storage capacity of the RAID 5 array is (n – 1) * c, where c
RAID 1 is 50%.
is the capacity of each of the disk drives. In this example, the total
storage capacity of the RAID 5 array is (6 – 1) * 18, or 90GB.
roring (or duplexing) two RAID 0 arrays. This yields the fault toler- is 1/n * 100, where n is the num-
ance of RAID 1 and the input/output speed of RAID 0. RAID 0/1 ber of disk drives in the array.
requires a minimum of four disk drives to implement. Figure 3.6
shows an example of a RAID 0/1 implementation.
RAID 0/1 (mirrored stripe sets without parity). Disk 1 Disk 2 Disk 3
RAID 1 mirror
Disk Controller
Stripe set [1,2,3]
is mirrored on
R E V I E W B R E A K
Which RAID level should you use? You must decide based on both
performance requirements and economic considerations.
RAID 0 provides excellent read and write performance, but provides
TIP
What Is RAID 0/1? Know that no fault tolerance. Because there is no fault tolerance in RAID 0,
RAID 0/1 combines the fault toler- you should not use it in a production network server.
EXAM
overhead for RAID 5 is equal to the capacity of one of the disk drives
in the array. For example, three 18GB disk drives in a RAID 5 array
have a storage capacity of 36GB.
Table 3.1 summarizes characteristics of the different RAID levels.
TABLE 3.1
CHARACTERISTICS OF RAID L E V E L S
Know Characteristics of the
TIP
RAID Level Minimum Number Read Write
Various RAID Levels Know the
of Drives Performance Performance
characteristics of the different RAID
EXAM
RAID 0 2 Excellent (but Very good (but levels presented in Table 3.1.
no fault no fault
tolerance) tolerance)
RAID 1 2 Very good Good
RAID 0/1 4 Excellent Very good
File Systems
A file system defines how files are stored on a disk drive. The original
file system used by early versions of the Disk Operating System
(DOS) was called the File Allocation Table (FAT) file system. The
original FAT used 12 bits to address clusters and was limited to a
08 Serv+ Ch_03 7/10/01 9:50 AM Page 208
What Is a Cluster? A disk cluster is the number of bits from 12 to 16 and could address a disk partition
a group of disk sectors (512 bytes) up to 2GB in size. This version of the FAT file system is now called
that are accessed together. This FAT16, to indicate that it uses 16 bits to address the clusters on the
means that a disk cluster is the
disk drive. Fat16 was used by early versions of Windows 95. The
smallest amount of disk space that
OSR2 (OEM Service Release 2) update to Windows 95 introduced a
can be allocated to a file. Common
new version of the FAT file system that uses 32 bits to address the
cluster sizes are 2KB (4 sectors),
4KB (8 sectors), 8KB (16 sectors), clusters on the disk drive, hence the name FAT32. FAT32 can
16KB (32 sectors), and 32KB (64 address large disk drives with ease. Using the default cluster size of
sectors). The larger the cluster size, 4KB, FAT32 can access a disk drive up to 8GB in size.
the more inefficient the use of the
The various network operating systems have their own file systems.
disk space. For example, a file con-
The following sections give some details about the various file
taining one character would be allo-
system that can be used by NOSs.
cated 8KB of disk space when the
cluster size is 8KB.
Windows NT Server 4
Microsoft’s Windows NT Server 4 supports both the FAT16 file
system and its own native file system known as NTFS (NT File
System). (Since the introduction of Windows 2000, the NTFS that
is used on Windows NT Server 4 has been called NTFS version 4.)
Installing Windows NT Service Pack 4 or higher allows Windows
NT 4 to read and write FAT32 file systems as well. However,
Windows NT Server 4 requires either FAT16 or NTFS on the disk
partition from which it boots.
The installation of Windows NT Server 4 itself was often done on a
FAT16 file system. This allowed a failed (and unbootable) network
NOTE
HPFS on Windows NT Previous ver- server running Windows NT Server 4 to be booted from a floppy
sions of Microsoft’s Windows NT (ver- disk using DOS and its FAT-based file system. Once booted from
sions 3.1, 3.5, and 3.51) supported the floppy, the files (including the NT system files) that were
the OS/2 HPFS file system. installed on the FAT16-formatted C: drive could be read and writ-
ten using standard DOS FAT16 utilities. This means that corrupted
Windows NT Server 4 system files stored on FAT16 could be
replaced from the original Windows NT Server 4 CD-ROM.
The Windows NT Server 4 file systems are as follows:
á FAT16. Partitions up to 4GB. (Partitions over 2GB are not read-
able by DOS.) Supports long filenames (up to 254 characters).
08 Serv+ Ch_03 7/10/01 9:50 AM Page 209
NOTE
4KB. Partitions up to 2 terabytes can be supported using larger What Is Unicode? Unicode is a 16-
cluster sizes. Supports long filenames (up to 254 characters). bit code that can represent 216, or
65,536 characters. This is more than
á NTFS version 4. File and volumes up to 16 exabytes (264 bytes, enough to represent all the characters
or 18,446,744,073,709,551,616 bytes). Supports long filenames in all the different languages currently
(up to 254 characters). Supports compression and NTFS file used on the earth.
system permissions. Stores characters and filenames as Unicode.
The root volume is C:\. All volumes on Windows NT Server 4 are
assigned drive letters (A–Z).
Novell NetWare 5
Even with the latest version of Novell’s NetWare, a small (Novell
recommends 50MB) FAT partition is required to start up the
NetWare server.
Novell NetWare supports two different native file systems: the origi-
nal NetWare File System (NWFS) and the newer Novell Storage
Services (NSS) file system.
The NWFS was available on previous versions of NetWare. Maximum
file size is 2GB. It has a maximum of 64 volumes, and each volume
can be 1TB in size. Supports file compression. The required SYS vol-
ume must use NWFS. Used on all previous versions of NetWare.
08 Serv+ Ch_03 7/10/01 9:50 AM Page 211
SYS:
|
| | | | | | |
ETC LOGIN MAIL SYSTEM PUBLIC NOVDOCS DELETED.SAV
UNIX
Each of the various versions of UNIX that run on an Intel-based
network server has its own, often unique, file systems. However,
most of them do support the FAT file system for compatibility with
other operating systems.
The file structure of UNIX/Linux systems differs significantly from
the more familiar DOS file structure used by many other network
operating systems, where each volume is assigned a drive letter. The
08 Serv+ Ch_03 7/10/01 9:50 AM Page 212
/ (aka root)
|
| | | | | | | |
/bin /dev /etc /home /lib /tmp /usr /var
Linux
The most popular native Linux file system is the ext2 (the second
TIP
What Is the Root of the extended file system) file system, which is an enhancement of an
UNIX/Linux File Structure? Know older extended file system known as ext. Other much less commonly
EXAM
that the name of the root of the used native Linux file systems include minix and xia. Minix is the
UNIX/Linux file structure is “/.” oldest of the Linux file systems and probably the least used due to its
lack of features (for instance, file names limited to 30 characters and
a maximum file system size of 64MB). Xia was an enhancement that
removed some of the file system restrictions of minix.
It appears that the Linux support community wants to be compatible
with every file system ever created; so for compatibility with other
operating systems, Linux also supports (among others) the following
(non-native) file systems:
á Msdos. For compatibility with FAT file systems
08 Serv+ Ch_03 7/10/01 9:50 AM Page 213
Server name
Password for the administrative account
If using TCP/IP protocol:
IP domain name
IP address
Subnet mask
Default gateway address
Address of DNS server(s)
Address of WINS server(s) (Microsoft networks only)
If using IPX/SPX protocol:
Frame type
You can use the following commands to view or change the network
configuration for various network operating systems:
Windows NT/Windows 2000 Server—ipconfig
Novell NetWare 4.x/5.x—inetcfg
UNIX/Linux—ifconfig
IBM OS/2 Warp Server—ifconfig
CONFIGURING EXTERNAL
PERIPHERALS
. 2.4 Configure external peripherals (UPS, external drive sub-
systems, etc.)
• Configure an external disk subsystem
• Configure an external CD-ROM system
• Configure an Uninterruptible Power Supply (UPS)
This section discusses the configuration of external disk subsystems,
external CD-ROM systems, and uninterruptible power supplies.
External disk subsystems are necessary when the amount of disk stor-
age cannot be accommodated by the disk drive bays internal to the
network server chassis. These external disk subsystems can be either
SCSI or Fibre Channel. Generally, Fibre Channel–based systems can
support many more disk drives than a SCSI-based external system.
08 Serv+ Ch_03 7/10/01 9:50 AM Page 217
In all cases, be sure that the power switch on the external disk sub-
system is turned on before turning on the network server.
What Are LUNs? Know that LUNs This means that you could have 7 CD-ROM drives all with the
allow multiple devices to share the SCSI ID of 5, each having a different LUN (1 through 7) all on the
same SCSI ID.
same SCSI channel. This means that on a single SCSI channel with
SCSI IDs of 1 through 7, you could have 7 LUNs for each SCSI
ID, for a total of 49 CD-ROM drives on a single SCSI channel.
To configure the external CD-ROM system, follow the manufac-
turer’s installation and configuration instructions. Be sure that the
external CD-ROM system is powered up before powering up the
network server.
Configuring an Uninterruptible
Power Supply
The installation of a UPS was covered in Chapter 2,
“1.0—Installation.” Be aware that a network server does
not automatically monitor a UPS. You must install UPS
monitoring software on the network server to give it this
capability. The network server needs to know when a power
failure has occurred and that the UPS is supplying power to
the network from its batteries. Follow the UPS vendor’s instruc-
tions to install the UPS monitoring software.
08 Serv+ Ch_03 7/10/01 9:50 AM Page 219
After you have installed the UPS monitoring software, you must
configure the following parameters:
á Time to wait before sending a warning to clients that the
server is running on battery power. This is generally a few
seconds. This allows time for power to be restored before send-
ing a message. This prevents warning messages from being sent
to clients when there is a momentary loss of power.
á Time to wait before beginning a shutdown of the network
server. This is generally a few minutes.
á Name of a program or group of commands to run as part
of the shutdown process. This may be just the name of the
server shutdown command, but it also may include programs
that do things such as sending a message to the network
server’s administrator’s pager along with the network server
shutdown command.
The UPS monitoring software often enables you to check the
status of the UPS. Status items include such things as the voltage
level entering the UPS as well as the voltage level leaving the UPS.
Figure 3.7 shows an example of the Windows 2000 Server UPS
Configuration dialog box.
FIGURE 3.7
UPS Configuration dialog box in Windows
2000 Server.
08 Serv+ Ch_03 7/10/01 9:50 AM Page 220
CONFIGURING MANUFACTURER-
SPECIFIC DRIVERS
. 2.6 Update manufacturer specific drivers
Always check the version number of drivers for devices in the net-
work server. Check the vendor’s web site for the latest driver for the
hardware device. You can always find the newest drivers available at
the hardware vendor’s web site. For example, check Compaq’s web
site (www.compaq.com) for the latest drivers for Compaq hardware
components. If the driver available on the manufacturer’s web site is
a newer version than the one you have installed on the network
server, download the newer version and install it. Record the driver
version for each hardware device in the network server as part of the
configuration documentation.
08 Serv+ Ch_03 7/10/01 9:50 AM Page 221
SNMP Community Know the with an SNMP community name. The default SNMP community
purpose of the SNMP community name is Public. You should always change this to something other
name. than Public. The community name is a primitive password mecha-
nism. SNMP management console and SNMP agents with the same
SNMP community name can exchange information. If the SNMP
community names differ, they will not be able to communicate.
Figure 3.8 shows the configuration of an SNMP community name
in Microsoft Windows 2000 Server.
FIGURE 3.8
Configuring SNMP community in Windows
2000 Server.
08 Serv+ Ch_03 7/10/01 9:50 AM Page 223
TIP
Each unique network device has a unique MIB defined. For exam- Base Know the purpose of the
SNMP MIB.
ple, there are MIBs for routers, hubs, switches, computers, and even
EXAM
some software packages, such as database systems. The MIB contains
a list of the items that can be managed by SNMP on the device
associated with the MIB.
TIP
The backup software for a network server has to be specific to the pose of backup software.
network operating system installed on the network server. For exam-
EXAM
ple, backup software for Microsoft Windows NT Server must be
certified to work with Microsoft Windows NT Server. There are var-
ious reasons for this. In the case of Windows NT Server, it is
required because the Windows NT Server Registry must be backed
up. Only backup software certified to work with Microsoft
Windows NT Server can back up the Registry. The Registry is a spe-
cial system file that is always open when Microsoft Windows NT
Server is running. Usually, backup software cannot back up open
files (files being used by other programs).
Backup software certified to work with Microsoft Windows 2000
Server can back up the “system state.” The system state contains
the Windows 2000 Registry; and on domain controllers, it also
contains the Active Directory and the contents of the system
volume (SYSVOL).
Some network operating systems provide a rudimentary backup
system. Unfortunately, this backup system is often not feature rich.
This lack of features often requires the purchase of a third-party
backup software package. To install the backup software, follow
the backup software vendor’s installation instructions.
pose of antivirus software. fers somewhat from the antivirus software that you would use on a
desktop computer. The biggest difference is that it generally costs
EXAM
purpose of system monitor agents. Major network server hardware vendors often have server monitor-
ing and management software available. To enable this software to
EXAM
Some network server vendors also offer optional hardware and soft-
ware that enable you to perform remote management of network
servers. The following are a few examples of the remote network
management hardware and software available from various vendors:
Compaq—Compaq Remote Insight Board
Dell—Dell OpenManage Remote Assistant Card (DRAC)
HP—HP Toptools Remote Control Card
HP—Integrated HP Remote Assistant
IBM—Advanced System Management Processor
FIGURE 3.10
Windows NT Server 4 paging file dialog box.
08 Serv+ Ch_03 7/10/01 9:50 AM Page 227
TIP
Performance Options button/Virtual Memory Change button. Paging Files Know the name of
the Windows NT Server 4 and
EXAM
Figure 3.11 shows the Windows 2000 Server paging file configura- Windows 2000 Server paging files
tion dialog box. and how to configure them.
FIGURE 3.11
Windows 2000 Server paging file dialog box.
Linux
Linux uses swap partitions, which are created by the installation
process and/or the administrator.
Adding NetWare Swap Files Know
TIP
The mkswap utility “formats” the swap partition. that NetWare allows one swap file
per volume. Also know that the
The swapon utility makes the swap partition(s) available.
EXAM
FIGURE 3.12
Microsoft System Monitor showing heavy
processor utilization.
08 Serv+ Ch_03 7/10/01 9:50 AM Page 230
CHAPTER SUMMARY
This chapter covered the job dimension having to do with the con-
KEY TERMS
figuring of the server. It dealt with the configuration of RAID on
• Antivirus software the network server, the selection of the proper file system, and the
• Backup software installation of the network operating system. It also covered the
• BIOS installation of common network server software, such as antivirus
• Cluster software and backup software. The chapter concluded by reminding
you to document all of these items.
• Disk array
• Disk duplexing
• Disk mirroring
• FAT
08 Serv+ Ch_03 7/10/01 9:50 AM Page 231
CHAPTER SUMMARY
• FAT16
• FAT32
• Flashable BIOS
• Hot spare
• HPFS
• LUN
• MIB
• NTFS
• NWFS
• NSS
• Paging file
• Partition
• Plug and Play
• RAID controller BIOS
• RAID 0
• RAID 1
• RAID 5
• RAID 0+1
• RAID 0/1
• RAID 10
• SCSI controller BIOS
• Sector
• SNMP
• SNMP community
• Stripe set with parity
• Stripe set without parity
• Swap file
• System BIOS
• Unicode
• Virtual memory
08 Serv+ Ch_03 7/10/01 9:50 AM Page 232
A P P LY Y O U R K N O W L E D G E
Exercises • Sun Solaris 8 Intel Platform
www.sun.com/software/solaris/
binaries/get.html
3.1 Installing and Configuring Network
Operating Systems • Linux versions
Installing a network server operating system is a major RedHat Linux (free, unsupported)
task. However, the experience gained by installing a www.redhat.com/download/
computer system that meets the minimum require- Caldera Linux (60-day trial and installa-
ments of the network operating system. The computer tion computer-based training)
system can be a desktop computer system for practicing www.calderasystems.com/products/
the network operating system installation. You also volution/cbt/
need a copy of the network operating system software
itself. If you don’t have a copy of the network operating After you have the computer system and copies of the
system software, you can find a free or low-cost copy (it network operating systems on hand, just follow the
may be a time-limited evaluation copy) of popular installation instructions of the network operating
network operating systems at the following web sites. system. Most network operating systems come on
You may want to get several different network operat- bootable CD-ROMs. As long as the computer system
ing systems so that you can practice the different instal- has a bootable CD-ROM drive, the installation process
lation scenarios. begins by inserting the network operating system
Estimated Time: 30 minutes to several hours (will vary CD-ROM into the CD-ROM drive and rebooting
with operating system). the computer system.
A P P LY Y O U R K N O W L E D G E
7. What is the name of and the default size of the 2. Which of the following technologies allows a net-
swap file on Microsoft Windows 2000 Server? work server to run more programs than will fit
into its physical RAM?
8. How many swap files are allowed in Microsoft
Windows 2000 Server? A. RAID 1
9. What is the purpose of UPS monitoring software? B. RAID 5
10. What is the name of the utility that enables you C. Virtual memory
to test TCP/IP network connectivity?
D. Backup software
11. What is the minimum number of disk drives
3. Which of the following software packages are not
needed to implement RAID 0/1?
normally installed on a network server to support
12. What is the most popular native Linux file system? its operation?
13. What processor statistics should be included in A. Antivirus software
the server baseline measurement?
B. Backup software
14. What is a SCSI LUN?
C. UPS monitoring software
15. What is the “top” of the UNIX/Linux file struc-
D. Spreadsheet software
ture called?
4. Which RAID level does not provide fault tolerance?
A. RAID 0
Exam Questions B. RAID 1
1. You have a group of six 36GB disk drives that you C. RAID 5
want to configure as a RAID 5 array. After you
D. RAID 10
have configured the RAID array with the six disk
drives, what is the total storage capacity of the 5. Which of the following is not a network server
single logical drive created by the RAID array? component that should be monitored to deter-
mine resource usage?
A. 216GB
A. Processor(s)
B. 180GB
B. Memory
C. 108GB
C. Power supplies
D. 36GB
D. Disk subsystem
08 Serv+ Ch_03 7/10/01 9:50 AM Page 234
A P P LY Y O U R K N O W L E D G E
6. What is the best place to find the latest driver 10. What is the name of the Windows NT Server 4
for a network server component? command that will display the current TCP/IP
configuration?
A. The network operating system vendor’s
web site A. winipcfg
B. The CD-ROM that was packaged with the B. inetcfg
hardware component
C. ifconfig
C. The hardware vendor’s web site
D. ipconfig
D. The network operating system CD-ROM
11. Where is the best place to look for the latest
7. You have four 18GB disk drives that you want security patches for a network server operating
to configure using RAID 0/1. What will be the system?
storage capacity of the RAID 0/1 array after it
A. The network server hardware vendor’s
has been configured?
web site
A. 18GB
B. The network operating system vendor’s
B. 36GB web site
C. 54GB C. The CD-ROM that came with the network
server
D. 72GB
D. The CD-ROM that contains the network
8. You discover that the very top of the file structure
operating system
for a network server is “/”. What network server
operating system is installed on the network 12. What is the maximum disk partition size that can
server? be addressed by the FAT16 file system?
A. Windows NT Server 4 A. 32MB
B. Novell NetWare B. 2GB
C. IBM OS/2 Warp Server C. 8GB
D. UNIX/Linux D. 32GB
9. What is the name of the IBM OS/2 Warp Server 13. The High Performance File System(HPFS) is
swap file? native to which network operating system?
A. PAGEFILE.SYS A. IBM OS/2 Warp Server
B. SWAPPER.DAT B. Windows 2000 Server
C. SWAPPER.SYS C. Novell NetWare
D. PAGEFILE.DAT D. UNIX
08 Serv+ Ch_03 7/10/01 9:50 AM Page 235
A P P LY Y O U R K N O W L E D G E
14. What is the storage capacity of a RAID 0 array 4. RAID 0, disk striping without parity, is not fault
that contains three 18GB disk drives? tolerant. It is used to improve disk drive
input/output. For more information, see the sec-
A. 18GB
tion “RAID 0 Configuration.”
B. 36GB
5. The system BIOS version displays on the server
C. 54GB monitor when the server is powered up. For
D. 9GB more information, see the section “Checking
System BIOS.”
15. What is the default SNMP community name?
6. The SNMP community is a primitive form of
A. Public security for the SNMP protocol. To access infor-
B. Private mation through SNMP, the community name of
the SNMP agent and the SNMP manager must
C. SemiPrivate be the same. For more information, see the sec-
D. Open tion “Installing and Configuring SNMP.”
7. The swap file for Microsoft Windows 2000
Server is called PAGEFILE.SYS. The default size
of PAGEFILE.SYS is 1.5 times the amount of
Answers to Review Questions RAM in the server. (For instance, a server with
1. The minimum number of disk drives required to 512MB of RAM will have a PAGEFILE.SYS file
implement RAID 5 is three. For more informa- that is 768MB in size.) For more information, see
tion, see the section “RAID 5 Configuration.” the section “Configuring Swap File(s).”
2. The disk overhead associated with RAID 1 is 8. There can be one swap file per volume in
50%. For more information, see the section Microsoft Windows 2000 Server. For more
“RAID 1 Configuration (Mirroring)” or “RAID information, see the section “Configuring
1 Configuration (Duplexing).” Swap File(s).”
3. The cute answer to this question is one disk con- 9. UPS monitoring software allows the network
troller. In disk mirroring, both disk drives are server to determine when a power failure has
attached to a single disk controller. In disk occurred and that the UPS is supplying power to
duplexing, each disk drive is attached to a differ- the network server from its batteries. For more
ent disk controller. Hence, the difference between information, see the section “Configuring an
mirroring and duplexing is one disk controller. Uninterruptible Power Supply.”
For more information, see the section “RAID 1
10. The ping command is used to test TCP/IP net-
Configuration (Duplexing).”
work connectivity. For more information, see the
section “Verifying Network Connectivity.”
08 Serv+ Ch_03 7/10/01 9:50 AM Page 236
A P P LY Y O U R K N O W L E D G E
11. RAID 0/1 requires a minimum of four disk drives 4. A. RAID 0 (disk striping without parity) does
to implement. For more information, see the not provide any fault tolerance of disk drives. It
section “RAID 0/1 Configuration.” does provide improved disk input/output perfor-
mance. For more information, see the section
12. The most popular native Linux file system is ext2
“RAID 0 Configuration.”
(the second extended file system). For more
information, see the section “Linux.” 5. C. The power supply is not a network server
component that requires monitoring to deter-
13. When performing a server baseline measurement,
mine percentage of the resource being utilized.
the processor percent utilization and the proces-
For more information, see the section
sor queue length statistics should be recorded.
“Performing a Server Baseline.”
For more information, see the section
“Performing a Server Baseline.” 6. C. The best place to look for the latest driver for
a network server component is the component
14. A SCSI LUN is a Logical Unit Number that
vendor’s web site. For more information, see the
allows multiple SCSI devices to share a single
section “Configuring Manufacturer-Specific
SCSI ID. For more information, see the section
Drivers.”
“Configuring an External CD-ROM System.”
7. B. The storage capacity of a RAID 0/1 array
15. The “top” of the UNIX file structure is named
with four 18GB disk drives is 36GB. For
“/” or root. For more information, see the section
more information, see the section
“File Systems.”
“RAID 0/1 Configuration.”
8. D. The UNIX/Linux operating systems have a
Answers to Exam Questions file structure with the top designed as “/” or
root. For more information, see the section
1. B. The total storage capacity of the single logical “File Systems.”
disk drive created by the RAID 5 array is 180GB.
Using the formula (n – 1) * c, (6 – 1) * 36 = 180. 9. B. The name of the IBM OS/2 Warp Server swap
For more information, see the section “RAID 5 file is SWAPPER.DAT. For more information,
Configuration.” see the section “Configuring Swap File(s).”
2. C. Virtual memory allows a network server to 10. D. The Windows NT Server 4 command that
run more programs than will fit in its physical will display the current TCP/IP configuration is
RAM. For more information, see the section ipconfig. For more information, see the section
“Configuring Swap File(s).” “Configuring the Network.”
3. D. Spreadsheet software is not used to support 11. B. The best place to look for the latest security
the operation of the network server. Spreadsheet patches for a network server operating system is
software may be installed on the network server the network operating system vendor’s web site.
for delivery to network client computers. For For more information, see the section “Installing
more information, see the section “Installing Network Operating System Updates.”
Service Tools.”
08 Serv+ Ch_03 7/10/01 9:50 AM Page 237
A P P LY Y O U R K N O W L E D G E
12. B. The maximum disk partition that can be 14. C. The storage capacity of a RAID 0 array with
addressed by the FAT16 file system is 2GB. For three 18GB disk drives is 54GB. For more infor-
more information, see the section “File Systems.” mation, see the section “RAID 0 Configuration.”
13. A. The High Performance File System (HPFS) is 15. A. The default SNMP community name is
native to the IBM OS/2 Warp Server network Public. For more information, see the section
operating system. For more information, see the “Installing and Configuring SNMP.”
section “File Systems.”
OBJECTIVES
4
ment upgrade
. Every server hardware specialist should be able to
add processors to a network server and upgrade the
C H A P T E R
network operating system to recognize multiple
processors.
OBJECTIVES
OBJECTIVES OUTLINE
OUTLINE S T U DY S T R AT E G I E S
Upgrading SCSI Hard Disk Drives 258 To prepare for the Upgrading objective of the
Adding SCSI Hard Disk Drives 258 Server+ exam, do the following:
Replacing SCSI Hard Disk Drives 258
. Read the objectives at the beginning of this
Adding Drives to a RAID Array 259 chapter.
New Drives in a Separate Array 259
New Drives in an Existing Array 259 . Study the information in this chapter.
. Review the objectives again.
Adding Memory 260 . Answer the Review and Exam Questions at the
end of the chapter and check your results.
Check Existing Memory 260
Checking Memory Upgrade Feasibility 261 . Use the ExamGear test on the CD-ROM that
accompanies this book for additional Exam
Checking Memory Upgrade Compatibility 262
Questions concerning this material.
Installing Additional Memory 262
. Review the Exam Tips available in various places
in this chapter. Make certain that you under-
Upgrading the BIOS/firmware 263 stand the information in the Exam Tip. If you
don’t fully understand the Exam Tip, review the
information in the chapter related to this topic.
Upgrading Adapters 265
Upgrading Adapter Memory 265
Upgrading Adapter BIOS or Firmware 266
Replacing an Adapter 266
INTRODUCTION
This job dimension covers the upgrading of network server hardware
and software and is the topic of approximately 12% of the exam
questions. This chapter covers the upgrading of various types of
network server hardware and software. The network server hardware
upgrades that are discussed include processors, hard disk drives,
memory, UPS, adapters, and peripherals. The software upgrades
discussed include BIOS/firmware, system monitoring agents, and
service tools.
Scheduling Downtime
Because most upgrades require that the network server be shut down
for a while, you need to schedule a time to perform the upgrade that
will have the least effect on the work being performed on the net-
work server. Unfortunately, this often means that many upgrades are
performed in the wee hours of the morning or on weekends.
Of course, the network server administrator should be involved in
the discussion of when to perform the upgrade. The downtime
should be announced so that everyone is aware that the network
server will be unavailable.
This material is likely to be asked You should adhere to the following list of best practices to ensure
about on the exam that you do not have a problem with ESD while working on a
EXAM
network server:
á Always use a static discharge device (static wrist strap, antista-
tic mat) when you open the server chassis to access compo-
nents in the server.
á Always store components in static bags until you are ready to
use them.
á Before you touch anything inside a network server, touch the
housing of the power supply. This discharges any static charge
that may have built up on your body.
á Use antistatic solutions on carpets in the server room.
á Use a room humidifier if the humidity is constantly below
50%.
á If possible, install antistatic carpet in the server room.
á Hold disk drives by the edges. Don’t touch the circuit board
on the bottom of the disk drive.
NOTE
SCSI disk drive additions—are verifiable during the power-up POST Items to Include in a Baseline You
of the network server. Other upgrades require that new drivers for the should include the following items in
hardware be installed after the hardware upgrade has been performed. your baseline measurement (as also
discussed in Chapter 2):
Some upgrades, such as migrating from a single processor network
server to a multiple-processor network server, may require major • Processor: Percent utilization
operating system software changes on the network server. • Processor: Processor queue
length
After you have completed the upgrade, and verified that both the • Disk subsystem: Percent disk
utilization
network server hardware and the network server operating system
have recognized the upgrade, you should review the upgrade process. • Disk subsystem: Disk queue
Note any major problems, because you might need to perform the length
same upgrade on another network server sometime in the future. • Network: Percent network
A should establish a new baseline measurement. You can compare utilization
this baseline to the original baseline measurement that was taken • Network: Network queue length
when the network server was originally installed. This comparison
should confirm whether the upgrade meets the intended goal of
increased performance.
As a final step, you should document the entire upgrade process in the steps outlined in the upgrade
network server log. Actually, you should be documenting the entire checklist.
EXAM
You also should verify the backup. During verification, you check
the contents of the tape against that of the corresponding files on
the disk drives. Performing a verification of the backup is usually an
option that you can select as part of the backup process. Verification
effectively doubles the time that it takes to perform a backup, but
you get the assurance that the contents of the disk drives and the
backup tapes match exactly.
ADDING PROCESSORS
. 3.2 Add Processors
• On single processor upgrade, verify compatibility
• Verify N+1 stepping
• Verify speed and cache matching
• Perform BIOS upgrade
• Perform OS upgrade to support multiprocessors
• Perform upgrade checklist
Processor upgrades fall into two general categories: replacing an
existing processor with a faster processor, or adding an additional
processor to a multiprocessor-capable network server.
STEP BY STEP
4.1 Performing a Single Processor Upgrade
Upgrading a Single Processor
TIP
1. Follow the upgrade checklist. Know the steps necessary to
2. Upgrade the system BIOS. upgrade a single processor to a
EXAM
faster processor.
3. Open the network server chassis (following ESD best
practices).
4. Remove the current processor.
5. Insert the new processor.
6. Close the network server chassis.
7. Verify that the new processor is recognized by the network
server hardware and the network operating system.
Figure 4.1 shows sample output from the Windows version of the
Intel identification program.
You can download the bootable floppy disk version of the Intel iden-
tification utility from http://support.intel.com/support/
processors/tools/frequencyid/bootable.htm.
FIGURE 4.1
Intel identification program (Windows version).
09 Serv+ Ch_04 7/10/01 9:51 AM Page 251
Figure 4.2 shows sample output from the bootable floppy disk
version of the Intel identification program.
You can find a description of the Intel identification program along
with installation instructions at http://support.intel.com/
support/processors/tools/frequencyid/download.htm.
You also can identify the current processor by a tag on the processor
itself. This tag contains a 5-digit S-spec number that starts with the
letter “S.” You can use this 5-digit number and the Intel web site to
identify the Intel processor that is in your network server. Figure 4.3
shows a sample Intel tag and identifies the S-spec number.
Frequency Test
CPUID Data
Utility completed. Eject the Bootable Intel(R) Processor Frequency ID FIGURE 4.2
Utilty version 3.0 floppy disk, then press "Ctrl + Alt + Delete"
Output from the bootable version of Intel
to restart the system.
identification program.
09 Serv+ Ch_04 7/10/01 9:51 AM Page 252
500V512/100/2.0V S1 You can use the S-spec number from the processor label and the fol-
FFFFFFFF-NNNN XXXXX
TM
©’98 SYYYY
lowing Intel web sites to identify the processor currently in your net-
work server:
S-spec
á Pentium
FIGURE 4.3 http://developer.intel.com/design/pentium/qit/
Intel processor S-spec number location.
á Pentium Pro
http://developer.intel.com/design/pro/qit/index.htm
á Pentium II
http://support.intel.com/support/processors/sspec/p2p.htm
á Pentium II Xeon
http://support.intel.com/support/processors/sspec/
p2xp.htm
á Pentium III
http://support.intel.com/support/processors/sspec/p3p.htm
á Pentium 4
http://support.intel.com/support/processors/sspec/p4p.htm
The processor that you are adding to the network server should
come with installation instructions. The installation instructions for
Intel processors are also available at the Intel web site.
You can obtain Intel processor installation manuals at
http://support.intel.com/support/processors/manuals/.
STEP BY STEP
4.2 Adding an Additional Processor
1. Follow the upgrade checklist.
2. Open the network server chassis (following ESD best
practices).
3. Insert the new processor(s).
4. Close network server chassis.
5. Upgrade the system BIOS.
09 Serv+ Ch_04 7/10/01 9:51 AM Page 253
TIP
Multiple Processors Know the steps necessary to
upgrade to multiple processors.
EXAM
Quite often a network server operating system will not recognize
that an additional processor has been installed on the network
server. The following sections detail how to remedy this situation.
STEP BY STEP
4.3 Activating Multiple Processor Support on
Windows 2000 Server
1. Click Start, Settings, Control Panel, and then click
System.
2. Click the Hardware tab, and select Device Manager.
3. Double-click to expand the Computer branch. Note the
type of support that you currently have.
4. Double-click the computer type that is listed under the
Computer branch, and then select the Drivers tab, Update
Driver, and then click Next.
5. Click Display a List of Known Drivers for This Device,
and then click Show All Hardware of This Device Class.
6. Click the appropriate computer type (a computer type
that matches your current type, except for multiple
CPUs), click Next, and then click Finish.
Novell NetWare 5
If your original installation of Novell NetWare 5 was on a network
server with a single processor, you must make several changes in the
configuration of the Novell NetWare 5 server for it to recognize and
use multiple processors. To upgrade Novell NetWare 5 so that it will
recognize the additional processor, follow these steps:
1. Load NWCONFIG | Multi CPU Options | Select a Platform
Support Module.
09 Serv+ Ch_04 7/10/01 9:51 AM Page 255
Disk System Upgrades Are existing network server, or replacing existing disk drives with larger
Dangerous to Data Before attempt- or faster disk drives. Upgrades to disk drives have the most potential
ing any upgrade to the disk subsys- of any upgrade to destroy data. Before attempting any disk drive
tem of a network server, make upgrade, make absolutely sure that you have at least one, preferably
absolutely sure that you have at least
two, verified full backups of the data on the disk drives.
one, preferably two, verified full back-
ups of the data on the disk drives.
á You can attach the second disk drive to the secondary ATA
channel using a second ribbon cable for the ATA controller.
You would set this single drive to either the single drive or
master ATA disk role depending on the manufacturer’s instruc-
tions for configuring a single disk drive on an ATA channel.
Chapter 2 details how to select the hard disk drive role. By putting
the second ATA disk drive onto the secondary channel, thus having
one ATA disk drive on each ATA channel, you can enhance the per-
formance of the disk subsystem.
A system BIOS upgrade might be required when upgrading from
small ATA disk drives to very large ATA disk drives. See “Upgrading
NOTE
ATA Interoperability Even though
BIOS/Firmware” later in this chapter for details on how to upgrade ATA is an ANSI standard, on occasion
the system BIOS. Older system BIOSs might not have the capability ATA implementations by different disk
to address all the space on very large ATA disk drives. The definition drive vendors don’t interoperate cor-
of “very large” has changed over the years. Various definitions of rectly. Be sure to check the disk drive
“large” have been 504MB, 1GB, 2GB, 4GB, and 8.4GB. Many of vendor’s web site for any known
issues between disk drives made by
these have proven to be barriers to ATA that had to be overcome by
different manufacturers. You can
newer, better, or improved BIOSs.
avoid conflicts by acquiring disk drives
Disk drive speeds are characterized by their rotational speed. from the same manufacturer, because
Common rotational speeds are 5400rpm, 7200rpm, and most manufacturers perform tests to
10,000rpm. The faster the rotational speed, the faster the disk can make sure that their own products
access data. Upgrading from slow disk drives to faster disk drives operate together correctly.
involves a complete replacement of the disk drives.
and you install two new disk drives that you want to configure into EIDE-Based RAID Upgrade Adding
a RAID 5 array using all five disk drives, for instance, you must ini- disk drives to an EIDE-based RAID
tialize all five disk drives. You can then combine the five disk drives array is similar to just adding disk
drives to a EIDE controller. You must
into a single RAID 5 array using all five disk drives. However, the
use the RAID configuration utility to
data that was on the original RAID 5 array (with the three disk
add the disk drives to the RAID array.
drives) will be destroyed in the process and will need to be reloaded
from the backup tape. The RAID controller will have a single RAID
5 array consisting of five disk drives.
09 Serv+ Ch_04 7/10/01 9:51 AM Page 260
ADDING MEMORY
. 3.4 Increase memory
• Verify hardware and OS support for capacity increase
• Verify memory is on hardware/vendor compatibility list
• Verify memory compatibility (e.g., speed, brand, capacity,
EDO, ECC/non-ECC, SDRAM/RDRAM)
• Perform upgrade checklist
It has been said that there is no such thing as too much memory in
a server. Although in many cases this is true, there are a few excep-
tions. One, you can only put as much memory in the network server
as it was designed to contain. There is always a maximum amount
of memory that can be supported by the processors and/or the con-
trol chipsets of the motherboard of the network server. The other
exception is having more memory than the network server operating
system can utilize. You must keep both of these exceptions in mind
when considering a memory upgrade to a network server.
STEP 2
Rotate the SIMM into the
upright position, making sure
FIGURE 4.4 the clips at each end of the
SIMM memory module installation. SIMM socket click into position.
09 Serv+ Ch_04 7/10/01 9:51 AM Page 263
EPROMs are Different EPROMs manufacturer. However, most BIOSs manufactured after 1995 can
(Erasable Programmable Read Only usually be upgraded by performing a “flash” upgrade, because the
Memory) are erased via exposure to BIOS code is stored in an EEPROM (Electrically Erasable
ultraviolet light. They also require a
Programmable Read Only Memory) . This is a simple process of
special device to allow them to be
replacing the BIOS program that is stored in flash memory. A flash
reprogrammed after they have been
BIOS upgrade is basically a software upgrade and does not require
erased.
that the network server chassis be opened.
The flash upgrade process is dangerous in that the old BIOS code in
the EEPROM is first erased before the new BIOS code is written to
the EEPROM. A disruption or loss of electrical power to the net-
work server during the flash upgrade process could leave the code in
the EEPROM in an incomplete and invalid state. Without valid
code in the BIOS, the network server cannot boot.
As with all upgrades, be sure to follow the upgrade checklist.
Although the process of upgrading a BIOS is vendor specific, Step
by Step 4.4 lists the procedure you must generally follow.
STEP BY STEP
4.4 Performing a Flash BIOS Upgrade
1. Obtain the latest BIOS program from the vendor (gener-
ally from the vendor’s web site).
2. Follow the vendor’s instructions as you load the BIOS
upgrade program onto a floppy disk drive.
3. Shut down and boot the server from the floppy disk con-
taining the BIOS upgrade program and the latest BIOS.
4. Follow the onscreen instructions as you perform the BIOS
upgrade.
5. Never interrupt the flash BIOS upgrade process, because
it could result in a network server that cannot be booted.
09 Serv+ Ch_04 7/10/01 9:51 AM Page 265
TIP
cally replace the old BIOS chip. A BIOS upgrade that requires the process of performing a flash
BIOS upgrade.
replacement of the BIOS chip is a hardware upgrade and ESD best
EXAM
practices must be followed (because you must open the network
server chassis to perform this upgrade). You must obtain the new
BIOS chip from the system board manufacturer.
Upgrades to the BIOSs or firmware on certain adapters (SCSI
adapter, RAID controller, FC-AL HBA) follow a very similar proce-
dure to upgrades to the system BIOS.
UPGRADING ADAPTERS
. 3.6 Upgrade adapters (e.g., NICs, SCSI cards, RAID, etc.)
• Perform upgrade checklist
Adapter upgrades also fall into several categories: upgrading compo-
nents on the adapter, such a memory; upgrading the BIOS or
firmware on the adapter; and replacing the adapter with a newer,
faster, or more powerful adapter.
Replacing an Adapter
You generally replace adapters after they fail. The replacement proce-
dure generally follows several simple steps:
1. Power down the network server.
2. Remove the defective adapter.
3. Install the new adapter.
4. Power up the network server.
Obviously this process requires that the network server be taken out
of operation. This results in downtime and lost productivity.
However, a recent technology known as PCI hot plug (or PCI hot
swap) enables you to replace, upgrade, or add an adapter without
powering down the network server.
09 Serv+ Ch_04 7/10/01 9:51 AM Page 267
For PCI hot plug to work, the network server hardware, the adapter
drivers, and the network server operating system must be PCI hot
plug aware. The network server hardware allows power to be
removed from individual PCI slots and allows adapters to be
removed and inserted without the use of a screwdriver. On a PCI
bus that supports hot plug, a slot release lever replaces the use of a
screw to secure the adapter in the PCI slot.
STEP BY STEP
4.5 Using PCI Hot Plug to Add an Adapter
1. Open the network server chassis.
2. Open the slot release lever on an available PCI slot. This
removes power from the select PCI slot.
3. Install the adapter into the selected PCI slot.
4. Attach any necessary cables to the adapter.
5. Close the slot release lever to secure the adapter in the
PCI slot.
6. Press the PCI hot plug button. This reapplies power to
the PCI slot.
7. The network server operating system locates and loads the
appropriate device drivers for the adapter or prompts the
installer for the location of the appropriate device drivers.
8. Close the server chassis.
09 Serv+ Ch_04 7/10/01 9:51 AM Page 268
additional adapters (NICs, SCSI Upgrading disk drives (EIDE and SCSI) is covered in the “Add Hard
controllers, RAID controllers, and so Drives” section in this chapter. Replacing, upgrading, or adding inter-
EXAM
STEP BY STEP
4.6 Replacing an Uninterruptible Power Supply
1. Shut down the network server.
2. Power off the network server and other devices that are
connected to the UPS.
3. Turn off the old UPS.
4. Disconnect all power cords from the old UPS.
09 Serv+ Ch_04 7/10/01 9:51 AM Page 273
CHAPTER SUMMARY
This chapter covered upgrades to various components of the net-
KEY TERMS
work server. You also learned about the upgrade checklist. The
• BIOS upgrade checklist is a list of procedures that should always be fol-
• Buffer lowed before, during, and after an upgrade to the network server.
• Cache This chapter started with a full description of the upgrade checklist.
You should consult the upgrade checklist and follow it as a regular
• Diagnostic partition part of any upgrade to a network server. Next, the importance of
• DIMM always performing a full, verified backup before starting any upgrade
was discussed.
• DMI
Discussion of upgrades to the network server hardware included
• EEPROM processors, hard disk drives, memory, adapters, peripherals, and
• EPROM uninterruptible power supplies. Upgrades to network server software
including the system BIOS, firmware, system monitoring agents,
• Firmware and service tools were also discussed in this chapter.
• Flash BIOS upgrade
• Frequently Asked Questions (FAQs)
• HTTP
• PCI hot plug
• Plug and Play
• RIMM
• Service tools
• SIMM
• SNMP
• System monitoring agents
• Upgrade checklist
• UPS hot swap
09 Serv+ Ch_04 7/10/01 9:51 AM Page 275
A P P LY Y O U R K N O W L E D G E
Exercises Estimated Time: 10 minutes.
1. Using a web browser access the Kingston
4.1 Identifying an Intel Processor Memory web site at www.kingston.com.
In this exercise, you download the bootable floppy 2. Access the memory information for a Compaq
DOS version of the Intel Identification Program and Proliant 5000 6/200.
use it to identify the processor in your computer system.
3. Answer these questions about the memory con-
Estimated Time: 15 minutes. figuration of a Compaq Proliant 5000 6/200 as
1. Download the DOS version of the Intel provided by the Kingston Memory web site:
Processor Identification Program from A. What is the standard memory configuration?
http://support.intel.com/
B. What is the maximum memory configuration?
support/processors/tools/frequencyid/
bootable.htm. C. How many memory modules must be added
simultaneously to this network server? (Hint:
2. Follow Intel’s instructions to create a bootable
Click the More Info link.)
floppy disk with the Processor Identification
Program installed. D. What memory modules are used in this net-
work server?
3. Shut down your computer.
E. How many memory sockets (slots) are in the
4. Insert the floppy disk into the floppy disk drive.
server?
5. The computer will boot and automatically run
F. What is the processor found in this server?
the Intel Processor Identification Program.
G. What is the bus architecture of this server?
6. Record the information that displays onscreen.
H. What memory kits are available for this
7. Use the Intel Processor ID links listed above to
server?
find information about the processor in your
computer. 4. Now try that again for a different network
server. While still at the Kingston Memory web
site, access the memory information for a
4.2 Identifying Network Server Memory
Hewlett-Packard NetServer LPr 750/800/850.
The objective of this exercise is to use a memory ven-
5. Answer the questions posed in step 2, but this
dor’s web site to identify memory in specific network
time about the memory configuration of a
servers. This will familiarize you with the process of
Hewlett-Packard NetServer LPr 750/800/850 as
using a memory vendor’s web site as well as the infor-
provided by the Kingston Memory web site.
mation about memory that can be obtained from a
memory vendor’s web site.
09 Serv+ Ch_04 7/10/01 9:51 AM Page 276
A P P LY Y O U R K N O W L E D G E
Review Questions Exam Questions
1. What is the purpose of verifying a backup? 1. Where is the best place to locate the latest
updates to the network server operating system?
2. What are system monitoring agents?
A. The network server vendor’s web site.
3. Where is the best place to get the latest drivers
for network server hardware components? B. The network server vendor’s support
CD-ROM.
4. What is the upgrade checklist?
C. The network server operating system vendor’s
5. What types of memory modules are most likely
web site.
to be found in a network server?
D. The network server operating system vendor’s
6. What is a FAQ?
installation CD-ROM.
7. How do you determine whether the current
2. Where is the best place to locate the latest drivers
processor in a network server can be replaced
for a third-party NIC installed in a network
with a faster processor?
server?
8. What are the requirements to add an additional
A. The network server vendor’s web site.
processor to your current network server?
B. The CD-ROM that came with the NIC.
9. How do you identify the processor currently in
your network server? C. The network server operating system vendor’s
web site.
10. Why must the network server operating system
be upgraded after adding an additional processor D. The NIC vendor’s web site.
to a network server that initially had a single net-
3. What is the very first step of any upgrade to a
work server?
network server?
11. What are the requirements to add a SCSI disk
A. Shut down the server operating system.
drive to an existing SCSI chain?
B. Back up the entire contents of the network
12. How do you add disks to a RAID 5 array?
server.
13. How do you determine the maximum amount of
C. Power down the network server.
memory that can be installed in a given network
server? D. Remove the cover of the network server.
14. Under what circumstances would you want to
upgrade the BIOS on a network server?
15. What is PCI hot plug?
09 Serv+ Ch_04 7/10/01 9:51 AM Page 277
A P P LY Y O U R K N O W L E D G E
4. After installing a second processor in a network i. Level 2 cache
server, the network server boots up just fine and
ii. Processor clock speed
the second processor is detected by the system
BIOS. However, the network server operating iii. Date of manufacture
system does not recognize the fact that there is iv. S-spec number
now a second processor in the network server.
How do you correct the situation? A. i and ii
A P P LY Y O U R K N O W L E D G E
9. How do you upgrade an EIDE disk subsystem to A. Install the NIC into PCI bus number one.
a SCSI subsystem?
B. Install the NIC into PCI bus number two.
A. Remove all EIDE disk drives, cables, and
C. You cannot install a NIC and RAID con-
controllers. Add the SCSI controller, cables,
troller in the same network server.
and disk drives.
D. The NIC has its own special slot because this
B. Remove the EIDE controller and replace it
is a network server.
with a SCSI controller, using existing cables
and disk drives. 12. What problem are you most likely to encounter
when adding external SCSI disk drives to a SCSI
C. Remove the EIDE controller, add the SCSI
bus?
controller, and change the drive electronics
from EIDE to SCSI. A. No available SCSI IDs
D. Remove the EIDE disk drives, install the B. Exceeding the SCSI channel cable length
SCSI disk drives, and attach the SCSI disk C. Slow performance
drives to the EIDE controller.
D. Heat
10. Which of the following adapters usually contain
onboard memory that can be upgraded? 13. What are the proper combinations for installing
memory modules into memory slots?
i. Video adapter
i. Gold leads on memory modules, gold
ii. Sound card contacts in memory slots
iii. RAID controller ii. Tin leads on memory modules, tin
iv. SCSI controller contacts in memory slots
A. ii and iii iii. Gold leads on memory modules, tin
contacts in memory slots
B. i, ii, and iii
iv. Tin leads on memory modules, gold
C. i, iii, and iv
contacts in memory slots
D. ii only
A. iii and iv
11. Your network server has dual-peer PCI buses. The
B. iii only
RAID controller is installed in PCI bus number
one. You need to install a 100Mbps NIC into the C. iv only
network server. Where do you install the NIC? D. i and ii
09 Serv+ Ch_04 7/10/01 9:51 AM Page 279
A P P LY Y O U R K N O W L E D G E
14. Which of the following limits the amount of 2. System monitoring agents monitor critical com-
memory that can be utilized in a network server? ponents in the network server and report their
status to a monitoring console. For more infor-
A. The control chipset on the network server
mation, see the section “Upgrading System
motherboard
Monitoring Agents.”
B. The speed of the processor(s) in the network
3. The best place to find the latest drivers for net-
server
work server hardware components is the hard-
C. The speed of the PCI bus ware component vendor’s web site. For more
D. The width of the data bus information, see the section “Locating the Latest
Software.”
15. What is the process of replacing a SCSI disk
drive with an SCA connector? 4. The upgrade checklist is a list of procedures that
should be followed before, during, and after an
A. Remove the old SCSI disk drive and insert upgrade to a network server. For more informa-
the new SCSI disk drive. tion, see the section “The Upgrade Checklist.”
B. Remove the old SCSI disk drive and note its 5. Most network servers will have SIMM, DIMM,
SCSI ID. Set the SCSI ID on the new SCSI or RIMM memory modules installed. For more
disk drive to the same SCSI ID and insert the information, see the section “Checking Existing
new SCSI disk drive. Memory.”
C. Remove the old SCSI disk drive and note its 6. FAQs are Frequently Asked Questions. A FAQ is
SCSI ID. Set the SCSI ID on the new SCSI a compilation of questions that are often asked in
disk drive to be different from the SCSI ID of newsgroups or are often asked of technical sup-
the old disk drive. Insert the new SCSI disk port personnel. For more information, see the
drive. section “Reviewing Instructions and FAQs.”
D. Remove the old SCSI disk drive. Set the new 7. To determine whether the processor in a network
SCSI disk drive to CSEL. Insert the new server can be upgraded to a faster processor,
SCSI disk drive. check the network server system board vendor’s
web site. For more information, see the section
“Replacing a Single Processor with a Faster
Processor.”
Answers to Review Questions
1. The processing of verifying a backup ensures that
the information stored on the tape matches
exactly the information stored on the disk drives.
For more information, see the section
“Performing a Full Backup.”
09 Serv+ Ch_04 7/10/01 9:51 AM Page 280
A P P LY Y O U R K N O W L E D G E
8. The requirements to add an additional processor 12. Back up the entire system. Add the disk drives as
to a network server are as follows: you would add disk drives to a SCSI chain. Use
the RAID configuration utility for the network
The network server system board must be multi-
server to delete the old RAID configuration and
processor capable.
create a new RAID configuration using the old
The additional processor must be the same disk drives and the additional disk drives. Restore
model, have the same clock speed, the same size the entire system from the backup. For more
Level 2 cache, and be within one stepping of the information, see the section “New Drives in an
existing processor. For more information, see the Existing Array.”
section “Installing Additional Processors.”
13. To determine how much memory can be
9. To identify the processor currently in a network installed in a given network server, consult the
server, use the Intel Processor Identification network server documentation, the network
Utility. For more information, see the section server vendor’s web site, or a memory vendor’s
“Installing Additional Processors.” web site. For more information, see the section
10. If there was only one processor in the network “Adding Memory.”
server when the network operating system was 14. There are many reasons to upgrade a system
installed, the operating system installed support BIOS in a network server, such as adding support
for a single processor. The network operating sys- for new hardware including processors, large hard
tem will not automatically detect the additional disk drives, and new bootable devices. Known
processor once it is rebooted. The operating sys- BIOS bugs are also corrected by BIOS upgrades.
tem must be upgraded to recognize the additional For more information, see the section
processor. For more information, see the section “Upgrading BIOS/Firmware.”
“Upgrading the Operating System for Multiple
15. PCI hot plug technology enables you to replace,
Processors.”
upgrade, or add an adapter without powering
11. The additional SCSI disk drive should use the down the network server. For more information,
same signaling system as the existing SCSI disk see the section “Replacing an Adapter.”
drives. The additional SCSI disk drive should be
of the same type as the existing disk drives. There
must be an available connector on the SCSI cable
for the additional SCSI disk drive. The new disk Answers to Exam Questions
drive must have a SCSI ID that differs from the 1. C. The best place to get the latest updates to the
existing disk drives on the SCSI chain. For more network server’s operating system is the network
information, see the section “Adding SCSI Hard server operating system vendor’s web site. For
Disk Drives.” more information, see the section “Locating the
Latest Software.”
09 Serv+ Ch_04 7/10/01 9:51 AM Page 281
A P P LY Y O U R K N O W L E D G E
2. D. The best place to get the latest drivers for a 9. A. To “upgrade” from an EIDE disk subsystem to
third-party NIC is the NIC vendor’s web site. For a SCSI subsystem, you must remove all the EIDE
more information, see the section “Upgrading disk drives, cables, and controllers, and replace
Peripheral Devices, Internal and External.” them with SCSI disk drives, cables, and con-
trollers. It isn’t really an upgrade, it is a complete
3. B. The very first step of any upgrade to a net-
replacement. For more information, see the sec-
work server is to fully back up the entire server.
tion “Upgrading IDE/ATA/EIDE/ATA-2 Disk
For more information, see the section
Drives to SCSI Disk Drives.”
“Performing a Full Backup.”
10. C. Video adapters, SCSI controllers, and RAID
4. C. The network operating system must be
controllers usually have onboard memory that
upgraded or reconfigured for it to recognize the
can be upgraded. For more information, see the
additional processor. For more information, see
section “Upgrading Adapter Memory.”
the section “Upgrading the Operating System for
Multiple Processors.” 11. B. The NIC should be installed on PCI bus
number 2. This will balance the load because the
5. A. The environment variable in Microsoft
RAID controller is on PCI bus number 1. For
Windows NT and Microsoft Windows 2000 that
more information, see the section “Upgrading
is set to the number of processors in the network
Peripheral Devices, Internal and External.”
server is Number_of_processors. For more infor-
mation, see the section “Microsoft Windows NT 12. B. A common problem encountered when
Server 4” or “Microsoft Windows 2000 Server.” adding external SCSI devices to a SCSI bus is
exceeding the maximum cable length for the
6. A. The processor cache size and the processor
SCSI channel. For more information, see the
clock speed must be the same on the two proces-
section “Adding SCSI Hard Disk Drives.”
sors in a network server. For more information,
see the section “Installing Additional Processors” 13. D. Don’t mix metals. Gold to gold. Tin to tin.
For more information, see the section “Checking
7. C. RIMMs require that continuity modules be
Memory Upgrade Compatibility.”
inserted into all unfilled memory slots in a net-
work server. For more information, see the sec- 14. A. The control chipset on the network server
tion “Installing Additional Memory.” motherboard can limit the amount of memory
that can be utilized in the network server. For
8. B. The second EIDE disk drive on the channel
more information, see the section “Adding
must be configured as a slave because the first
Memory.”
disk drive is configured as a master. For more
information, see the section “Upgrading 15. A. The SCA connector sets the SCSI ID. Just
EIDE/ATA-2 Hard Disk Drives.” remove the old SCSI disk drive and install the
new SCSI disk drive. For more information, see
the section “Replacing SCSI Hard Disk Drives.”
09 Serv+ Ch_04 7/10/01 9:51 AM Page 282
1. Mueller, Scott. Upgrading and Repairing PCs, 4. General computer hardware sites
12th Edition. Que Publishing, 2000. The PC Guide
www.pcguide.com
2. The Ultimate Memory Guide
Tom’s Hardware Guide
www.kingston.com/tools/umg/umg2000.pdf
www.tomshardware.com
3. Memory manufacturer’s web sites Whatis.com
www.kingston.com www.whatis.com
www.crucial.com Webopedia.com
www.micron.com www.webopedia.com
Free Online Dictionary of Computing
http://foldoc.doc.ic.ac.uk/foldoc/
index.html
10 Serv+ Ch_05 7/11/01 10:03 AM Page 283
OBJECTIVES
OBJECTIVES OUTLINE
OUTLINE S T U DY S T R AT E G I E S
Establishing Remote Notification 313 To prepare for the Proactive Maintenance objectives
of the Server+ exam, do the following:
. Read the objectives at the beginning of
Chapter Summary 315
this chapter.
. Study the information in this chapter.
Apply Your Knowledge 316 . Review the objectives again.
. Answer the Review and Exam Questions at the
end of the chapter and check your results
. Use the ExamGear test engine on the CD-ROM
that accompanies this book for additional Exam
Questions concerning this material.
. Review the Exam Tips available in various
places in this chapter. Make certain that you
understand the information in the Exams Tip. If
you don’t fully understand the tip, review the
information in the chapter related to this topic.
. Recognize the importance of doing backups on
a regular basis.
. Understand the different types of backups,
including
• Full
• Incremental
• Differential
. Create a baseline of performance on a system
and note how it changes over time and with
modifications to the system.
. Utilize SNMP (Simple Network Management
Protocol) to monitor TCP/IP performance on
your network.
. Be cognizant of physical issues regarding your
network and server(s). Pay attention to ambient
issues that can negatively affect performance,
such as air temperature.
. Understand how to verify the hardware installed
within a system.
. Realize the possibilities that exist for remotely
notifying an administrator when problems occur.
10 Serv+ Ch_05 7/10/01 9:53 AM Page 286
INTRODUCTION
Proactive maintenance is one of the most overlooked aspects of any
network. Too often an administrator must turn his attention to
putting out brush fires—at the cost of ignoring issues that give the
appearance of being less pressing.
When users complain that they cannot access their files, or a manager
needs to have a new printer added to the network immediately, you
must respond. These actions, which solve an immediate problem, are
reactive in nature. Often they are isolated incidents or issues that can
be resolved expeditiously.
Proactive maintenance can be summed up as everything you do to a
network, not in response to a brush fire, to ensure the system will con-
tinue to perform in the future. This is the most important concept
within this job domain. If you understand this, it will help you under-
stand the objectives within the domain, and it will help you
immensely in your job. It also will help you in answering the practice
questions that appear at the end of this chapter and the questions on
the actual exam.
UNDERSTANDING BACKUPS
. 4.1 Perform regular backup
An old joke has it that everyone always had plans to do a full backup
the very day a system crashed; it just crashed a bit too early. As sad a
fact as it may be, every system has the potential to crash; therefore,
you should perform backups regularly on all your systems.
One could argue that the importance of backups has diminished
over the years as other technologies such as RAID have been added
to the server. RAID is a wonderful technology that allows the server
to continue operating even in the event of a hard drive failure.
What RAID cannot do, however, is enable you to bring up another
system to continue operations after a tornado completely destroys
the server. Likewise, RAID is of little use when lightning strikes
outside your office building, traverses the ground, and fries all the
hard drives within the system. Page after page could be filled with
10 Serv+ Ch_05 7/10/01 9:53 AM Page 287
NOTE
clear by now: Backups provide an insurance that goes beyond most What Else RAID Doesn’t Do A
other technologies. worthwhile point to consider: RAID
doesn’t help when a user accidentally
Before going any further, it is important to understand what a deletes a crucial file. It isn’t just
backup is truly intended for. Assume the hard drives on a server system crashes that make backups
crash. You can always bring the machine up again by replacing the so worthwhile.
hard drives and reinstalling the operating system and related service
packs. You also can modify the configuration files and have the
machine back up on the network within a relatively short time.
What you cannot do so easily, however, is recreate the data that was
stored on the server.
Although a backup is often thought of as being a representation of
all the files on the system—including operating system and configu-
ration files—its real purpose should be to save the unique data that
your associates have created. If a network driver file gets lost, you
can grab an exact copy from another system, web site, and so on. If
the file was a bid to do business with your largest customer, however,
you cannot copy that back from anywhere else other than a backup
copy. It is this data that has real (monetary) value and that you
should be most concerned about backing up to be able to recover it
should it get lost or corrupted.
Defining “Regular”
The objective states that you should know how to perform backups
on a regular basis. “Regular” is a relative word, based on the impor-
tance/value of the data the backup holds. If you do backups only
once a week, you are accepting the possibility of losing up to one
week’s worth of data—a cost that most businesses cannot afford.
If you do backups once a night, the maximum data loss that can
occur is equal to one day’s work. Many businesses are comfortable at
this level, and it can be considered the norm. For some businesses,
however, nightly backups represent too high a cost. Instead, such
businesses measure the span between backups in hours. You should
always consider the value of the data being created and the type of
business the enterprise you are working with is in when computing a
backup frequency and defining “regular.”
10 Serv+ Ch_05 7/10/01 9:53 AM Page 288
FIGURE 5.1
Only the last file of these three has the
archive bit on.
10 Serv+ Ch_05 7/10/01 9:53 AM Page 289
FIGURE 5.2
You can ascertain the value of the archive bit
through the graphical interface.
An incremental backup backs up every file on the system that has its
archive bit on, and then turns off that bit when it completes. By
sheer virtue of backing up only those files that have this bit on, an
incremental backup is getting only a subset of all the files on the
system and can finish more quickly than a full backup.
A differential backup backs up every file on the system that has its
NOTE
archive bit on, and does not change the value of the bit. In other Other Forms of Backup Many backup
words, if the bit was on when the backup started (which it would utilities include different choices as
need to be to be included in the backup), it will continue to be on well, which are modified versions of
these three. A daily backup, for
when the backup completes. Again, by backing up only those files
example, is a backup of all the files
that have the bit turned on, a differential backup is getting only a
created or changed during a one-day
subset of all the files on the system and is thus able to complete
period. A copy backup is just a copy of
operation in a shorter time than a full backup. a file to media. (Think of copying one
Table 5.1 summarizes the differences among these three backup types. file to a floppy.) A partial backup is just
a copy of all the files within specified
directories. For this exam, you should
TABLE 5.1 concentrate only on the three major
backup types.
BACKUP TYPES
Type Includes Only Turns Archive Typical Running
Files with Bit Off upon Time (1=Longest)
Archive Bit On Completion
Full No Yes 1
Incremental Yes Yes 2
Differential Yes No 3
10 Serv+ Ch_05 7/10/01 9:53 AM Page 290
Differential Backups Full backups files are backed up, and the archive bits on them turned off. Sunday
can be mixed with either incremental night, only the files changed/created during the day Sunday are
EXAM
TABLE 5.2
RESTORING FROM A F R I D AY C R A S H
Type of Backup Number of Tapes Typical Restoration
Needed Time (1=Longest)
Full only 1 3
Incremental 6 1
Differential 2 2
Backup Media
By definition, a backup is intended to enable you to recover your
data in the event of hard drive failure(s). By implication, the location
of the backup must be other than the drives that could fail. Creating
a backup of the data on a disk to another partition on the same disk
is a waste of time and does not truly constitute a backup at all. Even
backup to a different fixed drive in the same server could turn out to
be useless in the event of a lightning strike or some other calamity
affecting the whole machine.
Until somewhat recently, backups were always synonymous with tape
media. Tapes are a removable media ideal for backups by virtue of
having the following characteristics:
á They have a large capacity. With compression, the amount of
data they can contain increases.
10 Serv+ Ch_05 7/10/01 9:53 AM Page 292
á They are easily transported. The days of tapes being bulky and
unwieldy have passed; now they take up very little physical space.
á They are based on proven, reliable technology. Adhering to
standards, it is possible for data on a tape created on one
system to be restored on another, even though the second
machine has a different type of tape drive.
á They are reusable. A backup tape can be overwritten or
appended to many times before degradation becomes a factor.
á They are relatively inexpensive. This makes it possible to buy
several sets of tapes for rotation, thus providing an extra level
of insurance; if one set of tapes goes bad, you have another to
turn to.
á They can withstand a broad range of climate conditions.
TIP
You Can’t Go Wrong with Tapes Although tapes continue to be a common backup medium, it is not
Although all the other possibilities unusual to use other media as well. Some other possibilities include
CD-RW discs, DVD-RW, Zip disks, and removable hard drives.
EXAM
STEP BY STEP
5.1 Backing Up Windows NT 4.0 Server
1. Windows NT 4.0 can back up only to tape, so verify that
a tape device is connected. Click Start, Programs,
Administrative Tools, Backup. (You also can start it from
the command line with NTBACKUP.EXE.)
2. Click the Backup button on the toolbar, or choose
Operations, Backup.
10 Serv+ Ch_05 7/10/01 9:53 AM Page 295
FIGURE 5.3
The backup utility from Windows NT has been
greatly enhanced with Windows 2000.
10 Serv+ Ch_05 7/10/01 9:53 AM Page 296
STEP BY STEP
5.2 Backing Up Windows 2000 Server
1. Click Start, Programs, Accessories, System Tools, Backup.
2. Choose the Backup tab and select the files/directories that
you want to back up.
3. Specify a backup media (destination) for the backup, and
then click Start Backup.
Both the target file and source files can be paths (such as /dev/tape).
The options include
á c To create a new file
á d To compare contents and display the differences between
the target and source
á f To specify a file
á p To keep the permissions
á r To append to an existing file
á t To show the names of the files in the tar file
á u To add only files that are new
á v To run in Verbose mode
á x To extract files
á z To compresses files
to create a new tar file on the tape and do so in verbose mode, and
tar xf feb.tar
to find all files on the system starting with “kad” and copy them
beneath the /home/kad directory, creating all the needed sub-
directories in the process.
After you have created the initial baseline, use the same settings and
capture data on a regular basis to compare against it. Place this
information in a database and analyze possible changes that can be
extrapolated from your findings. Has the system become slower? Has
it become more responsive? Are there fewer/more lost packets? With
a good set of data, you can analyze performance, isolate trends, and
plan proactively for problems that can be slowly creeping up.
FIGURE 5.4
The System Monitor tool in Performance
enables you to monitor system resources in
Windows 2000.
10 Serv+ Ch_05 7/10/01 9:53 AM Page 300
hard drives or more RAM—you are SNMP is the Simple Network Management Protocol. Running at
making the old baseline no longer
port 161 by default, it is the only way to obtain true statistics of net-
reflective of the system. Whenever
work usage under TCP/IP. Installation and configuration of SNMP
you make significant changes, you
need to toss away the old baseline
was discussed in Chapter 2, “1.0—Installation”; the focus here will
and create a new one for the new supplement, rather than repeat, that information.
configuration. SNMP enables network administrators to remotely troubleshoot and
monitor hubs and routers. Much of SNMP is defined within RFCs
1157 and 1212, although many more RFCs discuss SNMP. SNMP
can be found, along with other RFCs, on various web sites,
including http://rs.internic.net. You also can perform a search
on SNMP or RFC and find more specific information related to a
specific part of SNMP—for example, on just Ethernet and SNMP.
Using SNMP, you can find out information about these remote
devices without having to physically be at the device itself. This can
be a very useful tool if understood and used properly. You can find a
wide variety of information about these devices, depending on the
device itself, of course. Some examples include the following:
á IP address of a router
á Number of open files
á Amount of hard drive space available
á Version number of a host (such as Windows NT)
Before you set up SNMP, you need the IP address or host names of
the systems that will either be the initiators or those that will respond
to the requests. Microsoft’s SNMP Service, for example, uses the reg-
ular Windows NT host name resolution, such as HOSTS, DNS,
WINS, and LMHOSTS. Therefore, if you are using one of these res-
olution methods, add the correct host name to IP address resolution
for the computers that you are setting up with SNMP.
The types of systems on which you can find data include the following:
á Mainframes
á Gateways and routers
10 Serv+ Ch_05 7/10/01 9:53 AM Page 301
TIP
to provide remote management. SNMP is a distributed system, you
can spread out the management of
An SNMP Service by Microsoft enables a machine running
EXAM
it in different locations so as not to
Windows NT to be able to transfer its current condition to a
overtax any one PC, and for multi-
computer running an SNMP management system. However, this is ple management functionality.
only the agent side, not the management tools. Various third-party
management utilities are available, including the following:
á IBM NetView
á Sun Net Manager
á Hewlett-Packard OpenView
Usually, you have only one management system running the SNMP
Service per group of hosts. This group is known as a community.
Sometimes, however, you may want to have more. The following list
identifies reasons why you might want to have more:
á You may want to have multiple management systems make
different queries to the same agents.
á There might be different management sites for one community.
á As the network grows and becomes more complex, you may
need to help differentiate certain aspects of your community.
trap is set up from the management side on the agent. But the man-
agement does not need to go to the agent to find out whether the
trap information has been tripped. The agent sends an alert to the
management system telling it that the event has occurred. Most of
the time, the agent is passive except in this one occasion.
MIB Structure
The namespace for MIB objects is hierarchical. It is structured in
this manner so that each manageable object can be assigned a
globally unique name. Certain organizations have the authority to
assign the namespace for parts of the tree design.
10 Serv+ Ch_05 7/10/01 9:53 AM Page 304
The MIB structure is similar to TCP/IP addresses. You get only one
address from the InterNIC and then subnet it according to your
needs. You do not have to contact them to inquire about each address
assignment. The same applies here. Organizations can assign names
without consulting an Internet authority for every specific assignment.
SNMP Communities
A community is a group of hosts running the SNMP Service to
which they all belong. These usually consist of at least one manage-
ment system and multiple agents. The idea is to logically organize
systems into organizational units for better network management.
Communities are called by a community name. This name is case
sensitive. The default community name is “public” and generally all
hosts belong to it. Also by default, all SNMP agents respond to any
request using the community name “public.” By using unique
community names, however, you can provide limited security and
segregation of hosts.
Agents do not accept requests nor respond to hosts that are not from
their configured community. Agents can be members of multiple
communities simultaneously, but they must be explicitly configured
as such. This enables them to respond to different SNMP managers
from various communities.
Security
There really is no established security with SNMP. The data is not
encrypted, and there is no setup to stop someone from accessing the
network, discovering the community names and addresses used, and
sending fake requests to agents.
A major reason most MIBs are read-only is so that unauthorized
changes cannot be made. The best security you can have is to use
unique community names. Choose Send Authentication Trap and
specify a Trap Destination, and stipulate Only Accept SNMP
Packets from These Hosts.
You also might set up traps that let you know whether the agents
receive requests from communities or addresses not specified. This
way, you can track down unauthorized SNMP activity.
10 Serv+ Ch_05 7/10/01 9:53 AM Page 305
TABLE 5.3
SECURITY OPTIONS FOR THE SNMP A G E N T
Parameter Description
Send Authentication Trap Sends information back to the trap
initiator responding that the trap failed.
This could be because of an incorrect
community name or because the host is
not specified for service.
Accepted Community Names When a manager sends a query, a
community name is included; this is a
list of community names that the agent
will respond to.
Accept SNMP Packets from Responds to any query from any
Any Host management system in any community.
Only Accept SNMP Packets from Responds only to the hosts listed.
These Hosts
SNMP Agent
In some cases you will configure other aspects of the SNMP agent.
These set the type of devices that you will monitor and who is
responsible for the system.
The options available on this screen are as follows:
á The contact name of the person you want to be alerted about
conditions on this station—generally, this is the user of the
computer.
á The location is a descriptive field for the computer to help
keep track of the system sending the alert.
10 Serv+ Ch_05 7/10/01 9:53 AM Page 306
Any errors with SNMP are recorded in the system log. The log
records any SNMP activity. Use Event Viewer to look at the errors
and to find the problem and possible solutions.
3. An SNMP agent gets the packet and puts it into its buffer.
a. The community name is checked for validity. If it is not
correct or is corrupted, the packet is rejected.
b. If the community name checks out, the agent checks to see
whether the originating host name or IP address is correct
as well. If not, it is thrown out.
c. The inquiry is then passed to the correct dynamic link library
(DLL) as described in the preceding section on MIBs.
d. The object identifier gets mapped to the specific applica-
tion programming interface (API) and that call gets made.
e. The DLL sends the data to the agent.
4. The SNMP packet is given to the SNMP manager with the
requested information.
VERIFYING HARDWARE
. 4.5 Perform hardware verification
One of the simplest objectives within the exam, hardware verifica-
tion merely implies identifying the hardware that you have. Suppose,
for example, that there is an old server sitting in the closet and you
are considering using it as a print server. Your memory tells you that
it is a Pentium 233 with 64MB of RAM. But how can you verify
that before using it?
Many operating systems can provide information through the tools
within them to help with this. Figure 5.5 shows an example of the
System Properties dialog box in Windows 2000 Server. Using the
General tab, you can identify that the processor is at least a Pentium
(Family 6) and there is 128MB of RAM.
To get more detailed information, you can resort to the Device
Manager, as shown in Figure 5.6.
Step by Step 5.4 details how to see how many COM ports are in the
system on a machine running Windows 2000 Server and what
resources they are using.
10 Serv+ Ch_05 7/10/01 9:53 AM Page 309
FIGURE 5.5
The General tab of the System Properties
dialog box offers some identification information
in Windows 2000.
FIGURE 5.6
The Device Manager provides specific informa-
tion about resources and associated drivers in
Windows 2000.
10 Serv+ Ch_05 7/10/01 9:53 AM Page 310
STEP BY STEP
5.4 Verify COM Ports in Windows 2000 Server
1. Right-click the My Computer icon on the desktop and
choose Properties from the pop-up menu.
2. Select the Hardware tab in the System Properties dialog box.
3. Click the Device Manager button.
4. Click the plus sign (+) beside Ports (COM & LPT) to
expand the list. The ports installed in the machine display.
5. Right-click Communications Port (COM1) and choose
Properties from the pop-up menu.
6. Select the Resources tab and you can see the IRQ and I/O
resources associated with the port.
Within Linux, the proc filesystem structure can provide you with
information about your system. For example, /proc/interrupts
identifies the currently used interrupt request lines on your system,
as shown in Figure 5.7.
CPU0
0: 1025278 XT-PIC timer
1: 496 XT-PIC keyboard
2: 0 XT-PIC cascade
3: 1303 XT-PIC 3c589_cs
8: 1 XT-PIC rtc
11: 41 XT-PIC aha152x
12: 3588 XT-PIC PS/2 Mouse
13: 1 XT-PIC fpu
14: 110060 XT-PIC ide0
NMI: 0
[root@localhost /proc]#
FIGURE 5.7
The /proc files in Linux can identify hardware
resources.
10 Serv+ Ch_05 7/10/01 9:53 AM Page 311
Step by Step 5.5 outlines how to use this tool to create the log file
and identify resources on a machine.
STEP BY STEP
5.5 Using the Windows NT Hardware Query Tool
1. Insert the Windows NT Workstation CD into the drive.
When the installation pop-up menu appears, choose the
Browse this CD button.
2. Open the Support folder.
3. Open the Hqtool folder.
4. Insert a blank, formatted disk in the system and double-
click the MAKEDISK.BAT icon. The boot disk is now
created.
5. Shut down the machine on which you want to query the
hardware.
6. Insert the disk into the machine and boot it. After all the
querying is done, the resources identified are shown
within the tool. When you exit the tool, all the informa-
tion is written to a file named NTHQ.TXT (on the
floppy disk) that you can reference at any time.
FIGURE 5.9
The Messenger Service in Windows NT and
2000 can send alerts.
10 Serv+ Ch_05 7/10/01 9:53 AM Page 315
CHAPTER SUMMARY
Proactive maintenance is not a difficult process. It amounts to using
KEY TERMS
a good deal of common sense and looking for the possibilities for
problems to occur before they do. Most importantly, however, it • Ambient conditions
involves taking steps to ensure that when a problem does occur you • Archive bit
will know about it and have the resources in place to correct it.
• Baseline
Backups enable you to recover data in the event of a hard drive or
system failure, and baselines give you a starting point against which • Differential backup
you can gauge system performance. SNMP thresholds can send • Full backup
remote verification and alerts, letting you know what is going on
• Incremental backup
with your network. Physical housekeeping and hardware verification
are used to identify what you have and keep it operating under the • Proactive maintenance
best conditions possible.
• SNMP trap
10 Serv+ Ch_05 7/10/01 9:53 AM Page 316
A P P LY Y O U R K N O W L E D G E
Exercises 5.2 Interacting with the Messenger service
Through the Graphical Interface
5.1 Interacting with the Messenger Service
The same exercises done in 5.1 is repeated using the
from the Command Line
graphical interface rather than the command line. In
The Windows-based operating systems require the this exercise, you look to see whether the service is
Messenger Service to be running for remote notifica- currently running, and then stop and start it.
tion to even be a possibility. In this exercise, you look
Estimated Time: 10 minutes.
to see whether the service is currently running, and
then stop and start it (all from the command line). 1. In Windows 2000, select Start, Programs,
Administrative Tools, Services.
Estimated Time: 10 minutes.
2. An alphabetic list of services appears. Move down
1. Open a command-line window. To do so in
the list until you see Messenger. It should be have
Windows NT and 2000, select Start, Run, and
a status of Started and a startup type of
enter cmd.
Automatic.
2. Enter the command
3. To stop the service, right-click it and choose
net start
Stop from the pop-up menu. A progress bar
This will show, in alphabetic order, all services appears as the service is being stopped. When it
currently running. The Messenger Service, by is no longer running, the progress bar disappears
default, is automatically started upon boot, so it and there is no entry in the Status column for
should appear in the list. this service.
3. To stop the service, use the command 4. To start the service, right-click it and choose
net stop messenger Start from the pop-up menu. A progress bar
appears as the service is being started. When it is
Two messages should display; the first indicating running, the progress bar disappears and the
that the service is stopping, and the second indi- word “Started” displays in the Status column for
cating when it has stopped. this service.
4. To start the service, use the command
net start messenger
Review Questions
Two messages should display; the first indicating
1. What is the most complete type of backup that
that the service is starting (with a series of periods
you can create? Why is it the most complete?
that tick off as seconds pass), and a second indi-
cating that it has successfully started.
10 Serv+ Ch_05 7/10/01 9:53 AM Page 317
A P P LY Y O U R K N O W L E D G E
2. Martin is creating a log file of the performance of 2. Which TCP/IP service can be used for remote
key operating system parameters on the server. notification?
He wants to safely store the file for two months
A. Messenger
and then use it to compare against the perfor-
mance of the system at that time. What is Martin B. SMTP
said to be making? C. POP
3. Which protocol is used in TCP/IP to enable you D. SNMP
to monitor network operations using agents?
3. John did a full backup on the server on Tuesday
4. Servers should be kept away from other devices morning and it is now Thursday evening. Which
that could generate EMI. What does the abbrevi- type of backup will take the longest to run
ation EMI stand for? tonight?
5. In Windows 2000, the System Properties dialog A. Full
box can be used to identify the processor from
which tab? B. Incremental
A P P LY Y O U R K N O W L E D G E
6. Windows NT 4.0 enables you to compress the 10. Which tool was included with Windows NT to
Registry to where it will fit on one disk and be enable you to verify hardware in a machine
stored offsite. What tool is used to create this disk? before installing the operating system?
A. ERD A. FDISK
B. Backup B. NTHQ
C. RDISK C. SYS
D. Services D. COM
7. Windows 2000 enables you to create the same
compression of the Registry onto a single floppy Answers to Review Questions
disk. What tool is used in this operating system
for creating such? 1. A full backup is the most complete backup that
can be made on a system because it includes
A. ERD every file. For more information, see the section
B. Backup “Understanding Backups.”
C. RDISK 2. Martin is said to be making a baseline. A baseline
is a gauge of present performance that can be
D. Services
compared against future performance. For more
8. Which filesystem within Linux can tell you what information, see the section “Working with
resources are currently in use? Baselines.”
A. proc 3. SNMP is the Simple Network Management
Protocol, and it is used to monitor performance
B. etc
on a TCP/IP network. For more information, see
C. boot the section “Working with SNMP.”
D. tmp 4. EMI is an abbreviation for electromagnetic inter-
9. At what port does the SNMP Service run by ference. It can be caused by any device using
default? electricity but is common with such things as
fluorescent lights, industrial air compressor
A. 80 motors, and the like. For more information, see
B. 110 the section “The Physical Environment.”
A P P LY Y O U R K N O W L E D G E
6. To receive remote notifications, the Messenger 4. B. The reason plenum-grade cable exists is to run
Service must be running. To receive remote noti- it within this space. A is incorrect as PVC is the
fications “Establishing Remote Notification.” most common cable, and specifically cannot be
run in this space. C is incorrect because any non-
plenum-grade cable would be strictly prohibited
Answers to Exam Questions from being run in this place. D is incorrect as
1. C. He must first restore the full backup done on plenum-grade cable can be run in this space. For
Sunday night, and then the incremental done on more information, see the section “The Physical
Monday and Tuesday nights, for a total of three Environment.”
tapes. All other choices are incorrect because they 5. C. Device Manager can be used to identify
misstate the number of tapes needed for this resources in use in Windows 2000. A is incorrect;
operation. For more information, see the section the proc filesystem does not exist in Windows
“Understanding Backups.” 2000, but is a feature of Linux. B is incorrect
2. D. The SNMP (Simple Network Management because NTHQ makes a bootable disk but the
Protocol) is a service available for TCP/IP to question asks for a way to do so within Windows
monitor the network and alert you to problems. 2000. D is incorrect because it is a nonexistent
A is incorrect because the Messenger Service command/tool. For more information, see the
exists within Windows-based operating systems section “Verifying Hardware.”
but not across all platforms capable of running 6. C. Known as the Error Recovery Disk (presented
TCP/IP. B is incorrect because SMTP (Simple as choice A), it is created using the RDISK.EXE
Mail Transfer Protocol) is used to route mail. C is utility in Windows NT. In Windows 2000, the
incorrect because POP (Post Office Protocol) is utility was removed and the ability to create the
used to receive mail and nothing more. For more disk was moved to the Backup utility (choice B).
information, see the section “Establishing The Services utility enables you to start and stop
Remote Notification.” services, but has no bearing on the creation of the
3. A. Regardless of when any backup has been done ERD disk. For more information, see the section
last, a full backup will always take longer than any “Backups Within Windows NT.”
other type to run because it copies every file. B 7. B. Known as the Error Recovery Disk (presented
and C are both incorrect; but given the scenario as choice A), it is created using the Backup utility
outlined, they would take an equal amount of in Windows 2000. In Windows NT, the RDISK
time because they both would backup all files utility (choice C ) was used for this purpose. The
created or changed since the full backup was Services utility enables you to start and stop ser-
completed. D is incorrect because it would get vices, but has no bearing on the creation of the
only the files changed one day and be the shortest ERD disk. For more information, see the section
of those choices offered here. For more informa- “Backups Within Windows 2000.”
tion, see the section “Understanding Backups.”
10 Serv+ Ch_05 7/10/01 9:53 AM Page 320
A P P LY Y O U R K N O W L E D G E
8. A. The proc filesystem is dynamically created to 10. B. The NTHQ (NT Hardware Query) tool was
report the devices in existence on a particular included with Windows NT to enable you to
host within Linux. The other filesystems listed as query all components of your system and verify
choices are all valid directories under Linux, but that NT will run correctly before doing an instal-
none of them would hold any information rele- lation. A is incorrect because FDISK is used to
vant to the question. For more information, see partition a hard drive. C is incorrect because this
the section “Verifying Hardware.” utility is used to place an operating system on a
hard drive. D is incorrect because this is not the
9. C. The SNMP Service runs at UDP port 161 by
name of a valid utility (but represents a com-
default. A is incorrect because this is the port
monly used extension on many utilities). For
used for the WWW service. B is incorrect
more information, see the section “Checking
because this is the port used by the email service
Values Outside the Operating System.”
POP3. D is incorrect because it is the port used
by FTP. For more information, see the section
“Working with SNMP.”
1. Dulaney, Emmett and Chris Hare. LPI 3. Maione, Dennis. MCSE Training Guide:
General Linux II Exam Cram. The Coriolis Windows NT Server 4. New Riders Publishing,
Group, 2001. Discusses the Linux operating 1999. Includes discussion of backups,
system and hardware verification. Performance Monitor, and so on.
2. Weiss, Martin and Emmett Dulaney. i-Net+ 4. Maione, Dennis. MCSE Training Guide:
Exam Cram. The Coriolis Group, 2000. Windows 2000 Server. New Riders Publishing,
Includes coverage of TCP/IP ports and their 2000. Includes discussion of backups,
relationship to the operating system(s). Performance Monitor, and so on.
11 Serv+ Ch_06 7/11/01 10:04 AM Page 321
OBJECTIVES
C H A P T E R 6
5.0—Environment
11 Serv+ Ch_06 7/10/01 9:54 AM Page 322
OBJECTIVES OUTLINE
S T U DY S T R AT E G I E S
To prepare for the Environment objectives of the . Look around your work environment for some of
Server+ exam, do the following: the physical security and environmental condi-
tions, and see how the concepts discussed in
. Read the objectives at the beginning of this
this chapter apply. If you don’t work in a place
chapter.
where you have access to a server environment,
. Study the information in this chapter. go to a library and check the security associated
. Review the objectives again. with the walk-up machines. Look for sprinklers,
. Complete the exercises at the end of the chapter. and think about the other factors discussed in
. Answer the Review and Exam Questions at the the chapter. If you are a student, a high school
end of the chapter and check your results. or college computing facility is another place to
apply some of the concepts surrounding physi-
. Use the ExamGear test engine on the CD-ROM cal security and environmental factors.
that accompanies this book for additional Exam
Questions concerning this material.
. Review the Exam Tips available in various places
in this chapter. Make certain that you under-
stand the information in the Exam Tip. If you
don’t fully understand the Exam Tip, review the
information in the chapter related to this topic.
11 Serv+ Ch_06 7/10/01 9:54 AM Page 324
INTRODUCTION
In this chapter, the discussion moves outside the server box, and
starts looking at some of the other important things about maintain-
ing a production server: the environment in which the server is
placed. Often overlooked, the operating environment is a critical
factor frequently left out of the planning when a server is deployed.
The environmental factors discussed in this chapter include temper-
ature, humidity, static discharge, power, fire, flood, and finally, the
state of the server itself and ways to measure it. This chapter also
discusses the physical security considerations you need to take into
account as you architect your computing environment.
While reading about each of the components of the computing
environment, you also will learn how they apply to the three most
common environments in which you will find servers. To fully
appreciate the environmental factors, you need be familiar with
these three environments:
á Open-office environment
á Server room
á Computer data center
Open-Office Environment
The open-office environment is probably the most common environ-
ment for small business and workgroup servers. In this environment
you typically find servers in either a cube or maybe under a desk in
someone’s office. Although certainly not an ideal environment, this
environment is useful for businesses without the resources or space to
provide a dedicated server room. It can also make supporting a server
more convenient for a busy administrator. In larger companies it is
not uncommon to find development servers in the developer’s cube,
allowing easy access while application development is underway.
Server Room
The server room is generally a room dedicated to computing equip-
ment. You will frequently find the server, network equipment, and
cabling (sometimes even phone equipment will share this room).
11 Serv+ Ch_06 7/10/01 9:54 AM Page 325
Generally this is a locked area, and it provides a more secure area for
storing a server than under someone’s desk as discussed with regard
to the preceding environment. You will generally find dedicated
server rooms in branch offices of larger companies or in companies
where there are enough people to require network cabling and
phone wiring. In this environment it is not unusual to find the
servers located on the floor or on a table or desk that is not in use.
ENVIRONMENTAL ISSUES
. 5.2 Recognize and report on server room environmental
issues (temperature, humidity/ESD/power surges,
back-up generator/fire suppression/flood considerations)
Now that you know the basic types of environments you may find
your servers in, you need to learn the environmental factors you
must be concerned about when installing and maintaining a server.
The first piece of information you need is the operating parameters
of your server. The next page presents a sample server specifications
document. This should be your guide while designing or verifying
an operating environment for your server. This information can
generally be found on the server vendor’s web site, as well as in the
documentation that comes with the server.
Various sections of the server specifications document are discussed
throughout the chapter, with references to the specifics for each of
the environmental sections.
11 Serv+ Ch_06 7/10/01 9:54 AM Page 326
Why Do I Need to Worry About This computing environment means. Now it’s time to discuss the specific
When I Install the Server? There factors you should understand as you install your server.
are two reasons you need to take
these factors into account before you
install the server. First, if your environ- Temperature
ment is not up to standard at the
time of installation, you run the risk of Just like many homes, two things control the temperature of the
damaging the server either before you computing environment. First, during cold weather, the heating sys-
are able to correct the issues, or tem maintains the temperature at a comfortable level. This is applic-
sometimes while you are correcting able to the open-office and server room environments. Enterprise
the issues. One example is air condi- data centers are usually air conditioned year round, due to the heat
tioning. If your air conditioning is not generated by the equipment. During hot weather, you need to use
adequate, you run the risk of burning
air conditioning to ensure the server remains under the maximum
up the server due to lack of cooling.
operating temperature.
You also have a potential problem if
you try to add air conditioning after Whenever you are implementing a new server, one of the things you
the server is installed to correct the should check is the number of BTUs (British Thermal Units) that
issue. If your environment needs air your server puts out (this is sometimes difficult to determine,
conditioning added, and holes need to because the BTUs are frequently left off the server specifications)
be cut for vents, if you (or the contrac-
and make sure that the aggregate number of BTUs can be cooled by
tors) are not careful you can end up
your air conditioning unit. If you will be rack mounting the server,
with a server full of drywall dust,
you should also check to make sure the rack is well ventilated and if
potentially shortening the useful life
of the server. Even small holes in dry- possible, contains some sort of airflow management (such as a fan)
wall can put an unbelievable amount to ensure the server stays cool when in a closed rack enclosure.
of dust into an enclosed area such as Liebert (www.liebert.com) is a leading manufacturer of precision air
an office or server room.
conditioning systems for use in critical environments such as data cen-
The second reason will probably not ters. Liebert’s enterprise air conditioners actually measure their cooling
be mentioned on the Server+ exam, capacity based on the volume of air in the data center. The Liebert
but is equally valid. If you install the web site is an excellent place to find more information on data center
server and it appears to be working environmental controls, as they are the industry leader in data center
fine, justifying the costs of correcting cooling systems, offering systems ranging from 6 tons to 60 tons.
environmental deficiencies to manage-
ment becomes that much more diffi- One of the easiest environmental issues to identify is the tempera-
cult. If your management happens to ture. If you walk into your computer room and it is really hot, you
have a technical background it may have a problem. In this case a thermometer on the wall can be your
not be as large an issue, but the “If best friend. One thing you should be very aware of is the designed
it’s not broken, don’t fix it” mentality operating temperature for your equipment. A typical server would
still persists in many companies. operate in the following ranges:
Operating temperature: 50º F to 95º F
Storage temperature: – 40º F to 105º F
10,000 BTU/hour max heat dissipation
11 Serv+ Ch_06 7/10/01 9:54 AM Page 329
In this example you can see that if the area your computer is stored
in exceeds 95 degrees, you will probably have a serious problem. How Much of This Do You Need to
TIP
Know for the Exam? Because
IN THE FIELD configuring and installing enterprise
EXAM
environmental conditioning equip-
MELTING DOWN COMPUTERS ment is almost always done after a
lot of consulting with the vendor, or
I once worked in a government building in an office area that also is done by someone with an exten-
housed some special-purpose computers. One of these was an old sive background in data center
machine used to gather and process a series of reports on system design, the details of the Liebert
accesses. It was a very old machine and could not be repaired or offerings (or comparable competitor
replaced. The year we were in the building summer came early to offerings) are beyond the scope for
New York with temperatures in the 90s in mid-May. The temperature the exam. It is good information to
in our office, with all the PCs and other computers, was in excess of have, but you won’t need to start
100 degrees during the middle of the day. And because this was a memorizing statistics.
controlled environment building, opening the windows was not an
option. We had some fans, but we really needed the air conditioning
turned up. The facilities manager informed us that it didn’t get hot
enough in New York for the air conditioning until June 10th, and there
was nothing he could do. The machine in question burnt itself up
about five days before the air conditioning came on. All the data and
programming was lost, and the reports were never generated again.
IN THE FIELD
RAISED FLOORS
continued
NOTE
Humidity
NOTE
How Big an Issue Is Humidity? It is
Humidity is another environmental quality that is critical to the very rare for humidity to be an issue
proper functioning of your server. As you can see in Table 6.1, the in a standard office setting. Where
relative humidity is another metric that is usually included in the you need to be very careful about
server specifications. Too much humidity in the air and the moisture humidity is in areas where the envi-
can damage your server’s electronic components. Too dry and your ronment either has a very high or
environment can be subject to electrostatic discharge (ESD) which very low humidity. If your server is
located in a greenhouse, you proba-
will be discussed later in this chapter in the “Electrostatic Discharge”
bly need to keep a close eye on the
section of the chapter.
humidity. If you are in a very dry, arid
environment where you notice a lot of
static, you will probably want to inves-
TABLE 6.1
tigate using a humidifier to bring the
C O M M O N E L E C T R O S TAT I C D I S C H A R G E humidity up into the server’s operat-
PREVENTION DEVICES ing specifications.
Device Description
Mats and flooring Floor mats, floor tiles, mat tape, bench and shelf mats.
Wearable protection Wrist straps, coil cords, ESD watches.
Floor grounding Foot grounders, sole grounders, and ankle straps.
ESD packing materials Bags, boxes, tape, trays, and cabinets.
Monitors and testers These devices are used to test the ESD devices to
ensure they are discharging appropriately.
Air Quality
NOTE
NOTE
power-line filter, which will filter an incoming power signal What if Your Equipment Is in a
to smooth out variations. Although not effective against sags, Smoking Area? Although smoking in
offices is becoming less of an issue
these can be very effective against spikes and surges and can
as the anti-smoking laws become
be a lifesaver in the event of catastrophic power surge, such as
more wide spread, if you have equip-
a lightning strike. It is generally better to replace a $25 surge ment in an area where smoking is
suppressor than a $5,000 server. You should evaluate two fac- permitted, expect problems. The parti-
tors when selecting a surge suppressor: cles in tobacco smoke are just as
• Clamping speed. The speed at which a surge suppressor dangerous (and in some cases more
dangerous) than other environmental
reacts to a power spike or surge. The faster the clamping
particle hazards. You should avoid
speed, the better the surge suppressor.
exposing your equipment to tobacco
• Clamping voltage. The voltage at which the surge suppres- smoke whenever possible.
sor reacts to a spike or surge. The lower the clamping volt-
age, the better the surge suppressor.
á Uninterruptible power supply (UPS). Uninterruptible power
NOTE
After the Strike Keep in mind that
supplies are battery-based systems designed to monitor incom- you can see just as much damage
ing power and then take over when an unacceptable voltage is done by the surge that occurs when
measured on the line. There are two basic types of UPS: power is restored after an outage as
you see caused by an outage. You
• Standby power system. A standby power system monitors
should always be sure to power off
the power line for an unacceptable power variation to and/or unplug any systems that have
occur. The battery in this model is not part of the power lost power due to an outage to avoid
line and draws current only to remain charged. In the damage (or additional damage) being
event of an interruption of power, the UPS switches from done when the power comes back on.
line power to battery power, converting the direct current
power stored in the battery to an alternating current volt-
age like the line power in use. This is done using an
inverter circuit, and the switch can take up to 10 minutes.
NOTE
also consider the amount of time the UPS can supply that
NOTE
What if the Power Goes Out When power. Ideally you should have an automated shutdown set to
No One Is in the Office? Most busi- occur in the event of a power outage; but if you do not, you
ness-quality UPSs can be purchased should ensure the server will operate long enough for you to
with software and cabling to allow
shut it down manually. Usually 15 to 30 minutes of operation
them to automatically shut down your
are adequate.
server in the event of an outage.
Make sure you test this process thor- Four standard 420W redundant hot-pluggable power supplies
oughly. The worst thing to find out the
Redundant hot-pluggable cooling fans
day after an outage is that your UPS
ran out of battery power one minute 110/220 volts
before the software completed the 200VAC single-phase @ 16A
shutdown.
3.1 kVA max consumption
IN THE FIELD
LINE CONDITIONING
This chapter has discussed the various types of protection for your
server power, but only discussed line conditioning briefly. This topic
is usually considered only when implementing a large-scale server
installation. Usually the answer for a server in an office or a server
room is to connect the server to a standby UPS. This is the most
cost-effective solution, and works very well in the event of a hard
outage. Because the battery and cutover circuits are not in line
with the power, however, the power-conditioning capabilities of this
11 Serv+ Ch_06 7/10/01 9:54 AM Page 335
type of UPS are limited. They will usually have circuits for handling
surges and spikes, but their capability to supplement power during
a brownout is limited. A true UPS, with its batteries in line, has the
capability to supplement power from the batteries during periods of
sags or brownouts. If you have a facility using a backup generator,
the odds are pretty good that you have additional line-conditioning
equipment designed specifically to supply your data center with a
clean line voltage.
Now that you know more about the important topic of power, con-
sider some of the other environmental factors you should take into
account when you are installing a server.
Fire Suppression
You may be asked to install servers in several different environments.
You need to be aware of fire-suppression considerations in these areas.
In a data center environment, you will generally have an inert gas-
based fire-suppression system that works by driving all the oxygen
out of the air. Halon was a favorite for many years but is being
phased out in favor of more environmentally friendly gases such as
carbon dioxide. Table 6.2 includes a list of some of the more com-
mon clean replacement chemicals for Halon.
TABLE 6.2
NOTE
What Do the Letters on the Fire reducing the oxygen content of the protected area below the point
Extinguisher Mean? The letters where it can support combustion. DuPont originally developed
have the following meanings: FE13 as a chemical refrigerant, and it suppresses fire by absorbing
A indicates fires where ordinary com- heat from a fire until the atmosphere can no longer support the fire.
bustibles such as wood, cloth, and
Another type of fire suppression is the use of foam to extinguish the
paper are burning.
fire. Foam fire-suppression systems work by separating the fire from
B indicates fires where flammable the oxygen in the air. Depending on the type of foam system, this
liquids, oils, and grease are burning. can be done by using the foam to blanket the fuel surface, smother-
C indicates fires where live electrical ing the fire by cooling the fuel with the water content of the foam,
equipment. (This is the one you want.) or by suppressing the release of flammable vapors that can mix with
the air and burn.
D indicates fires containing com-
bustible metals such as magnesium, In a server room you generally have the same sprinkler system you
potassium, and sodium. would find out in the office space. This is generally a remarkably bad
idea. Not only is water one of the worst things you can pour on an
electrical fire, but what if the fire is in a wastebasket on the other end
of the floor? Suddenly the sprinklers go off and your server resembles
the Fourth of July with all the sparks shooting out of it. Step one in a
server room should be to have maintenance disable the sprinkler. The
next step should be to buy a good ABC or BC fire extinguisher. (See
the sidebar for details on the lettering code.) Mount this prominently
either directly inside or on the wall outside the door to your server
room. If there is a fire in the server room, you should have ready
access to a fire extinguisher rated for electrical fires.
In a shared office environment, it is generally not practical to disable
the sprinklers. Your best bet in this instance is to try to shield the
server from any possible water from the sprinklers. This can mean
placing the server under a desk, enclosing the server in some sort of
sealed rack, or otherwise shielding it from moisture. You should keep
an ABC or BC fire extinguisher handy in this case as well.
Flooding
You should sincerely hope that you never have to deal with a flood-
ing situation with regard to your servers. In most cases if you are
unable to save the equipment before the flood, it will be damaged
or completely unrecoverable. In the event of impending flooding,
however, you can do a few things:
11 Serv+ Ch_06 7/10/01 9:54 AM Page 337
NOTE
bly not going to be dependable. If you want to try to recover How Big a Jolt Can You Get from
them, you should completely remove all components and dry ESD? The actual voltage involved
them as quickly as possible. Allow the components to air dry with ESD varies widely. Humans start
to feel a static shock at about 3,000
for at least 48 hours before cleaning and testing them.
volts. It is important to remember that
Corrosion is one area of concern and moisture on the compo-
although the voltage is high, the
nents is another. If you apply electric current to a motherboard amperage is very low, which is why
that still has a couple droplets of water on it, you could experi- ESD is does not injure people. Static
ence some “shocking” results. charges of up to 30,000 volts are not
uncommon and can be generated by
If you work in a known flood plain, make sure your flood insurance the simple act of walking across a
is paid up, or at least have an excellent disaster-recovery plan in carpeted floor. To damage a piece of
place in case of floods. sensitive electronics, a charge of as
little 10 volts can do the trick.
Electrostatic Discharge
Electrostatic discharge (ESD) is defined as the transfer of charge
What Should You Know About
TIP
Now that you are familiar with the environmental factors you
should consider when installing a server, it’s time to discuss securing
your server.
IN THE FIELD
IN THE FIELD
NOTE
4. Keep track of your keys. You should know how many keys have
been made for each lock, to whom they have been issued,
where the spares are stored, and whether the lock is on a build-
ing master key. You also need to ensure you know who has
access to that key, as well as ensure that there are effective key
control measures in effect for the building master keys. It is
also an excellent idea to make users sign for keys.
IN THE FIELD
NOTE
mon of the access control mechanisms, these readers work by Who Was Sleeping in My Computer
reading a magnetic signature off an access card or token. The Room? You would think that with the
ability to tell which card was used to
advantage of this system over a key system is you can authorize
access the computer room, you would
and remove access from the central authentication server. So
be able to tell who was in stealing
instead of having to rekey a lock and redistribute keys, or reset a your servers. Unfortunately that is not
combination and redistribute the combination, you just need to the case. Although you can tell whose
turn off the access card or token. This yields a much more card was used to gain access, you
secure environment than the previous mechanisms. You also have no way of telling who was hold-
have the added benefit of being able to track access based on ing the card at the time.
access token. You can tell what token was used to access an area.
IN THE FIELD
NOTE
What About Effective Card Control?
One company I used to work for paid an external security firm to The most effective way to ensure
perform a security audit of one of their facilities. The first thing the security using card readers is three-
security consultant did was ask someone in the hall who he fold. First, periodically review who has
needed to see to get a card for access to the building. When told been issued cards, and what areas
HR, he followed the signs and told the secretary in HR he was they have access to. Make sure they
working on a project and needed access to the building. The secre- are still appropriate. As an adjunct to
tary asked what parts of the building, he told her the entire build- that, also make sure that cards are
ing, and 5 minutes later badged himself into the data center. returned by departing employees, or
An uproar ensued, and procedures were promptly put in place to at least ensure their cards are
ensure that people would be authorized before receiving a card. deleted from the system after they
The next day the same consultant arrived in HR and requested card leave. Second, make sure no cards
access. The secretary, being aware of the new policy, asked who are issued without written authoriza-
had authorized his access. He gave her the name of the security tion. Finally, review usage. If you see
officer who had engaged him. Five minutes later he badged himself someone on the day shift is in the
into the data center again, with his shiny new access card. data center from 3:00 a.m. until 5:00
a.m. every night for a week, check
and make sure that he or she is
á Biometrics. Biometrics is the most promising access control supposed to be there.
technology on the horizon for future use. In fact some of these
products are available today, and as the cost comes down you
will see more and more of these systems in place. A biometrics-
based access control system uses measurable physical character-
istics to authenticate users into an environment. Examples of
biometrics that might be used to authenticate a user include
fingerprints, retinal patterns, or speech. There have even been
some discussions about using DNA or scent as the biometric
key for these systems in the future. Although the field is still in
11 Serv+ Ch_06 7/10/01 9:54 AM Page 344
What About the Server under My critical role in future computers, and especially in electronic
Desk? You may have noticed that commerce. There are some concerns from privacy advocacy
this discussion does not cover how to groups that these types of authentication systems are an inva-
secure a server in a shared office
sion of privacy.
environment. That’s because you
really can’t secure a server in that
All right, now that you have learned the best way to keep the general
environment. It is very difficult to
secure the space under your desk.
public away from your equipment, it’s time to discuss some of the
Some of the server-specific pre- other things you should be doing to physically secure your server.
cautions won’t hurt, but you should
be aware that your server just isn’t
secure until there is a locked door Establish Anti-Theft Devices
between it and the rest of the world.
for Hardware
Now that you have secured the perimeter, consider some of the security
measures you can take in case an intruder sneaks in through a window.
NOTE
IN THE FIELD
CHAPTER SUMMARY
This chapter has discussed a number of environmental factors that
KEY TERMS
could impact the heath and well being of your server, including the
• Biometrics following:
• Blackout á Temperature
• British thermal unit (BTU) á Humidity
• Brownout á Air quality
• Clamping speed á Power-line issues
• Clamping voltage á Fire suppression
• Electrostatic discharge (ESD) á Flooding
• Sag á Electrostatic discharge
• Spike á Server state
• Surge
This chapter discussed these environmental factors with an eye
• Transients toward three different types of server environments:
• Uninterruptible power supply (UPS) á Open-office environment. This environment typically has
• VA (volt-ampere) servers in either a cube or someone’s office, and is common in
smaller companies, or for development servers.
á Server room. The server room is a room dedicated to com-
puting equipment. You will generally find dedicated server
rooms in branch offices of larger companies, or in companies
where there are enough people to require network cabling and
phone wiring.
á Computer data center. A computer data center is the ideal
place to store your server. An environmentally controlled,
secure, raised-floor area specifically designed for storing large
numbers of servers, minicomputers, and mainframes—these
are expensive areas generally limited to larger companies.
The important thing to understand when you are designing your server
environment, or adding a new server to an existing environment, is
what the operating requirements for that server are. For this exercise,
the server’s specifications are your best friend. If you don’t understand
what type of environment your server needs to function, you will not
be able to complete your installation and ongoing support successfully.
11 Serv+ Ch_06 7/10/01 9:54 AM Page 349
CHAPTER SUMMARY
Next you learned some of the aspects of physical security. The first
important concept dealt with securing the computing environment.
The key methods include the following:
á Lock and key
á Combination locking mechanisms
á Card readers and proximity readers
á Biometrics
A P P LY Y O U R K N O W L E D G E
Exercises 9. Are all production servers secured within a locked
rack?
6.1 Complete a Physical Security Review 10. Are the cases on all production servers locked,
Checklist and is access to the keys controlled?
The purpose of this exercise is to get you thinking 11. Is removable media kept secured at all times?
about the factors involved in reviewing physical secu- Where is this media secured (desk, locked cabi-
rity issues. This checklist is a good basis for a physical net, next to the server, offsite)?
security review, but should in no way be considered
exhaustive. 12. How recently have the physical security measures
been audited? Was it an internal or external
Estimated Time: 30 minutes. audit? Were deficiencies revealed corrected?
In your computing environment, complete the follow- 13. If a card reader system is being used, how often is
ing checklist concerning the physical security of the usage monitored? By whom?
facility. If you do not have access to a computing
environment, you can use a public library, university,
or other public area that contains computers. 6.2 Complete an Environmental Review
Checklist
1. How are the premises protected against external
intruders (guards, cameras, fences, secure parking The purpose of this exercise is to get you thinking about
area)? the factors involved in checking for environmental issues.
This checklist is a good basis for an environmental
2. How is the building protected (security doors, review, but should in no way be considered exhaustive.
locked/barred windows, building guards)?
Estimated Time: 30 minutes.
3. How is access to sensitive areas within the build-
ing maintained (keys, combination locks, prox- In your computing environment, complete the follow-
imity readers)? ing checklist concerning the computing environment.
If you do not have access to a computing environment,
4. Who controls access to the sensitive areas of the you can use a public library, university, or other public
building (facilities, security, IT)? area that contains computers.
5. Is there a security alarm on the building? On the 1. Are all your servers located in an environmentally
sensitive areas within the building? controlled space (data center, air conditioned
6. How frequently is access to sensitive areas reviewed server room)?
(monthly, semi-annually, annually, never)? 2. Is adequate fire protection in place for the com-
7. Are key control/card control measures in place puting environment (fire suppression, building
for departing employees? firewalls, fireproof doors)?
8. Are all production servers secured within a con- 3. Are all servers on uninterruptible power supplies?
trolled access area? Do these UPSs provide enough power to ensure
the controlled shutdown of the server?
11 Serv+ Ch_06 7/10/01 9:54 AM Page 351
A P P LY Y O U R K N O W L E D G E
4. If the servers are located in a rack, are there fans 3. You’re the administrator of Little Faith Enterprise’s
in the rack to provide sufficient cooling? network. Your manager has asked you to install
three mission-critical mail servers in your location.
5. Are fire extinguishers of the correct type available
What environmental factors should you consider
for fire outbreaks outside of the computer room?
before installing these servers?
Is this a building sprinkler system, fire extin-
guishers, or both? 4. Where can you find the information you need to
determine whether your operating environment
6. Is the facility on a flood plain?
will support the server you want to deploy?
7. Are the computers stored in a facility with good
5. You’re a LAN administrator for Think About IT
air quality? Areas of concern include office
Consulting Services Company. You have noticed
environments where a lot of printing occurs,
that that the server backup tapes for your depart-
construction areas, manufacturing floors, and
ment’s server are stored in a cardboard box under
smoking areas.
the table the server is on. Why is this a problem,
8. Are measures in place to reduce or prevent ESD and how should you correct it?
(mats, static bags, static straps)?
9. Is the temperature monitored? How often?
10. Is the server’s operating state (temperature, line
Exam Questions
voltage) monitored? Are alerts generated on an 1. You are the network administrator for Wild
error condition (sag, temperature exceeds operat- Widgets Inc. and have been asked to install a badge
ing limits)? reader system to secure your computer data center.
Which of the following are advantages of a badge
reader system?
Review Questions i. Low cost.
1. You are the network administrator for Exponent
ii. Eliminates the need for key control.
Mathematicians and your manager has asked you
to “lock up those servers!” He recommends that iii. Allows for more flexible access management.
you have a lock installed on the server room and
iv. Enables you to track who accesses a spe-
distribute keys to the rest of your team. What are
cific area.
the benefits and drawbacks to this security
method? v. Badges are easier to distribute and main-
tain than keys.
2. What are the advantages and drawbacks to each
A. ii, iii, and v
of the remaining access control mechanisms dis-
cussed in the chapter? Which is the most appro- B. i and iv
priate for your environment? C. i only
D. iv only
11 Serv+ Ch_06 7/10/01 9:54 AM Page 352
A P P LY Y O U R K N O W L E D G E
2. You are the system administrator for Phil’s Phill-up 4. You are the LAN administrator for Little Faith
Stations, a chain of gas stations. As part of your Enterprises Meat Packing. Your manager wants to
responsibilities you have been asked to add surge be able to track who goes in and out of the meat
suppressors to all the servers to prevent an outage refrigerators to prevent theft.
in the event of a lightning strike or power surge.
Which technology do you choose?
What two factors should you consider when
A. Lock and key
selecting your surge suppressor?
B. Combination lock
i. Clamping speed
C. Badge reader
ii. Clamping voltage
D. Biometric system
iii. Amperage
5. You are the lead engineer for Little Faith
iv. Number of outlets
Enterprises. You are installing a server in your
v. Surge capacity data center. You notice that the data center is get-
ting very hot, and you need to keep your systems
A. i and iii
cool. It appears the air conditioner is failing.
B. i and ii What are some ways of cooling the system?
C. ii and iv i. Rent a portable air conditioner.
D. iv and v ii. Remove the cover of the server.
3. You are the LAN administrator for the OUI iii. Power off unused or unnecessary equipment.
Find-em detective agency. You are using a phone
iv. Open a window or door.
closet as your computer room, and your server
just caught on fire. Which of the following would A. i and ii
be appropriate for putting out the fire?
B. ii and iii
i. A BC-type fire extinguisher
C. iii and iv
ii. An AD-type fire extinguisher
D. i and iii
iii. An ABD-type fire extinguisher
6. You are the network administrator for the Hittem
iv. An ABC-type fire extinguisher Boxing Glove Corporation. You are installing
three new servers in your data center, and want to
v. A bucket of water
rack mount them.
A. i and ii
What features should you be looking for in a rack
B. ii and iii system to ensure the adequate cooling and good
C. i and iv physical security?
D. iv and v
11 Serv+ Ch_06 7/10/01 9:54 AM Page 353
A P P LY Y O U R K N O W L E D G E
i. Front and rear locking doors 8. You are the LAN administrator for UR Write
Publishing, a publishing company. You are
ii. Side panels
responsible for maintaining the data center envi-
iii. A KVM switch ronment. You have been tasked with providing
iv. Fans power fault tolerance for your mission-critical
data center.
v. Cable management
What is the best solution for providing power
A. i, iii, and v fault tolerance in a data center?
B. ii, iv, and v A. Surge suppressor
C. i, ii, and iv B. Battery backup
D. iii, iv, and v C. Uninterruptible power supply
7. You manage the servers for the Bang Bang D. Backup generator
Hammer Corporation. Your director is very con-
cerned about protecting the system from power 9. You are the collaborative computing administra-
problems since his neighbor’s house was hit by tor for the Call-me Telephone Company. You
lightning. have just installed your first UPS and are plug-
ging in devices.
What are the three types of power-line conditions
he needs to worry about? Which of the following should you plug into the
UPS?
i. Wrong amperage
i. Server
ii. Too many watts
ii. Monitor
iii. Line transients
iii. Router
iv. Low voltage
iv. Printer
v. Outage
v. Copier
A. i, ii, and iii
A. i and v
B. ii, iii, and iv
B. iv and v
C. iii, iv, and v
C. i, ii, and iii
D. i, iii, and v
D. iii, iv, and v
11 Serv+ Ch_06 7/10/01 9:54 AM Page 354
A P P LY Y O U R K N O W L E D G E
10. You are the server administrator for Jim’s House of • Card readers and proximity readers. The
Wrenches. Your main data center just had a pipe advantage of this system over a key system is
burst in the hallway outside your data center. Your you can authorize and remove access from the
data center is in danger of filling with water. central authentication server. You can also
track access based on access token. This is
What is the first thing you should do to protect
generally a more expensive solution, and you
your equipment?
have to deal with lost tokens.
A. Call your insurance company.
• Biometrics. The advantage to biometrics is
B. Unplug all the servers. there is no key control, because you are using
C. Remove all the hard drives from the servers. a person’s characteristics (fingerprints and so
forth) for access control. This is a very new
D. If your backup media is stored in the data technology, and is still very expensive.
center, move it to a dry location.
For more information, see the section “Physically
Securing the Data Center/Computer Room.”
A P P LY Y O U R K N O W L E D G E
Answers to Exam Questions 5. D. Adding additional cooling capabilities and
reducing the number of heat-generating devices
1. A. Using a badge reader removes the need for key are both excellent ideas. Removing the cover of
control, enables you to be more flexible in what the server actually can make the heat a bigger
areas people can access; and because badges can problem, because many servers incorporate the
be disabled if lost or stolen, they are also easier to cover into the cooling mechanism of the server.
distribute than keys. It is not a low-cost solution, Most data centers don’t have windows, due to the
so i is not the correct answer; iv is incorrect security issues, and it would really only help cool
because all you can tell is what badge was used to the room if it were winter, and you weren’t in a
enter the area, and not who physically used the hot part of the country. For more information,
badge, you can’t track the actual person who see the section “Temperature.”
accessed an area. For more information, see the
section “Physically Securing the Data 6. C. To secure the rack, you need locking front and
Center/Computer Room.” rear doors, and a fan will help keep the systems
cool. Although a KVM switch and cable manage-
2. B. When you are looking for a surge suppressor, ment are nice features, they don’t contribute to the
you really want to look at the clamping speed environment or security of the rack. For more
and voltage. The speed tells you how quickly the information, see the sections “Temperature” and
suppressor will respond to a voltage spike, and “Establish Anti-Theft Devices for Hardware.”
the clamping voltage tells you how much voltage
it will take to trigger the suppressor. The amper- 7. C. Transients, low voltage, and outage are the
age is not really a function of the surge suppres- three conditions to worry about. Wrong amper-
sion, and the number of outlets won’t help age and too many watts are not problems that
prevent an outage. The surge capacity is a occur. For more information, see the section
meaningless term. For more information, see “Power-Line Issues.”
the section “Power-Line Issues.” 8. D. Although all these provide some type of fault
3. C. BC or ABC are the types of fire extinguishers tolerance for power, the backup generator is the
that should be used in the event of a fire in your best solution for a data center environment. It is
server. An AD or ABD wouldn’t have the proper certainly not the least expensive. For more infor-
chemicals for combating an electrical fire. Water mation, see the section “Power-Line Issues.”
is always a bad idea with electrical equipment. 9. C. A server, monitor, and router are all
For more information, see the section “Fire relatively low-power, mission-critical devices.
Suppression.” Devices such as printers and copiers have such
4. D. You need a biometric system, because that is high power requirements that they will burn out
the only technology that enables you to track the a UPS. For more information, see the section
individual based on his/her physical characteristics. “Power-Line Issues.”
For more information, see the section “Physically
Securing the Data Center/Computer Room.”
11 Serv+ Ch_06 7/10/01 9:54 AM Page 356
A P P LY Y O U R K N O W L E D G E
10. D. Although A, B, and C are all possible answers,
your first concern should be protecting the data.
Get your tapes out of there. Then save whatever
servers you can, keeping in mind electricity and
water is a dangerous combination. For more
information, see the section “Flooding.”
1. Tyska, Louis A. and Lawrence J. Fennelly. 2. Toigo, Jon William. Disaster Recovery Planning,
Physical Security: 150 Things You Should Know, Prentice Hall PRT, 2000.
Butterworth-Heinemann, 2000.
12 Serv+ Ch_07 7/11/01 10:04 AM Page 357
OBJECTIVES
OBJECTIVES OUTLINE
OUTLINE
S T U DY S T R AT E G I E S
To prepare for the Troubleshooting and Problem . Use the ExamGear test on the CD-ROM that
Determination objectives of the Server+ exam, do accompanies this book for additional Exam
the following: Questions concerning this material.
. Read the objectives at the beginning of this . Review the Exam Tips available in various places
chapter. in this chapter. Make certain that you under-
. Study the information in this chapter. stand the information in the Exam Tip. If you
don’t fully understand the exam tip, review the
. Review the objectives again. information in the chapter related to this topic.
. Answer the Review and Exam Questions at the
end of the chapter and check your results
12 Serv+ Ch_07 7/10/01 9:56 AM Page 361
INTRODUCTION
This chapter covers the Troubleshooting and Problem Determination
(hardware and software) section of the Server+ exam. This job
dimension is the topic of approximately 27% of the exam questions.
As with any troubleshooting task, knowledge of how the system
works proves very useful in finding and eliminating the problem. The
same applies here; knowledge from the other chapters of this book is
very useful in troubleshooting problems with a network server.
Not only does this chapter cover troubleshooting techniques, it also
discusses ways to find network server bottlenecks. You will also learn
various ways to eliminate network server bottlenecks.
PERFORMING PROBLEM
DETERMINATION
. 6.1 Perform problem determination
• Use questioning techniques to determine what, how, when
• Identify contact(s) responsible for problem resolution
• Use senses to observe problem (e.g., smell of smoke, obser-
vation of unhooked cable, etc.)
Troubleshooting is often more of an art than a science. The ability
to be a good troubleshooter is frequently found in people who
always break everything into very small logical steps and then follow
those steps to a very logical conclusion. There are way too many
people whose approach to troubleshooting is more like firing a shot-
gun than a rifle. With the shotgun approach to troubleshooting,
many solutions are tried at one time in the hope that one of them
will solve the problem. This is not the correct approach to trou-
bleshooting. Effective troubleshooting is a methodical process, where
there is only one change made at a time.
12 Serv+ Ch_07 7/10/01 9:56 AM Page 362
Of course, you may also ask other questions, but this group should
cover most of the ground. In addition to asking questions of those
directly connected to the server’s operation, you can certainly benefit
from talking to other contacts as well.
to contact for specific information. To aid in troubleshooting some problems, the server hardware spe-
Have that information on hand and cialist should maintain a list of contact people who can offer assis-
EXAM
NOTE
Add to Tool Kit A server hardware
A good troubleshooter uses his or her senses to help find problems. specialist’s tool kit should include a
Questions that can be answered by senses include: digital thermometer. The digital ther-
mometer can be used to get an exact
Is the server room too hot? temperature reading for the server
Is the server room too humid (above the maximum operating room.
humidity for the equipment)?
Is there the smell of something burning in the server room?
Is smoke visible in the server room?
NOTE
FIGURE 7.1
Windows NT Task Manager Performance tab.
12 Serv+ Ch_07 7/10/01 9:56 AM Page 366
FIGURE 7.2
Windows NT Task Manager Processes tab.
FIGURE 7.3
Windows NT Task Manager Applications tab.
12 Serv+ Ch_07 7/10/01 9:56 AM Page 367
FIGURE 7.4
Windows NT Diagnostics.
12 Serv+ Ch_07 7/10/01 9:56 AM Page 368
FIGURE 7.5
Windows NT Event Viewer System log.
12 Serv+ Ch_07 7/10/01 9:56 AM Page 370
FIGURE 7.6
Windows NT Event Viewer Security log.
FIGURE 7.7
Windows NT Event Viewer Application log.
12 Serv+ Ch_07 7/10/01 9:56 AM Page 371
C:\>ipconfig /?
Windows NT IP Configuration
C:\>ipconfig
Windows NT IP Configuration
12 Serv+ Ch_07 7/10/01 9:56 AM Page 372
IP Address. . . . . . . . . : 10.20.30.40
Subnet Mask . . . . . . . . : 255.255.255.0
Default Gateway . . . . . . : 10.20.30.1
C:\>ipconfig /all
Windows NT IP Configuration
nbtstat
The Windows NT Server nbtstat command enables you to
NOTE
-a (adapter status) Lists the remote machine’s name table given its name.
-A (Adapter status) Lists the remote machine’s name table given its
➥IP address.
-c (cache) Lists the global remote name cache including the IP
➥addresses.
-C (cache) Lists the remote name cache with IP addresses on a
➥per-device basis.
-n (names) Lists local NetBIOS names.
-r (resolved) Lists names resolved by broadcast and via WINS.
-R (Reload) Purges and reloads the remote cache name table.
-S (Sessions) Lists sessions table with the destination IP addresses.
-s (sessions) Lists sessions table converting destination IP
➥addresses to host names via the HOSTS file.
-RR (ReleaseRefresh) Sends Name Release packets to WINs and then, starts
➥Refresh.
RemoteName Remote host machine name.
IP address Dotted-decimal representation of the IP address.
interval Redisplays selected statistics, pausing interval seconds
➥between each display. Press Ctrl+C to stop redisplaying
➥statistics.
C:\>nbtstat -n
netstat
The Windows NT Server netstat command enables you to display
information about current TCP/IP connections to the network
server. The flags for the Windows NT Server netstat command are
as follows:
C:\>netstat /?
C:\>netstat -s
IP Statistics
ICMP Statistics
Received Sent
Messages 34566 176
Errors 0 0
Destination Unreachable 34560 175
Time Exceeded 5 0
Parameter Problems 0 0
Source Quenchs 0 0
Redirects 0 0
Echos 1 0
Echo Replies 0 1
Timestamps 0 0
Timestamp Replies 0 0
Address Masks 0 0
Address Mask Replies 0 0
TCP Statistics
UDP Statistics
nslookup
The Windows NT Server nslookup command enables you to test
host name resolution by using a DNS server. The nslookup com-
mand can be used in two ways: command line or interactively.
The commands available in nslookup when it is used interactively
are as follows:
C:\>nslookup
Default Server: dsn.myisp.com
Address: 10.18.6.103
> help
Commands: (identifiers are shown in uppercase, [] means optional)
NAME - print info about the host/domain NAME using default server
NAME1 NAME2 - as above, but use NAME2 as server
help or ? - print info on common commands
set OPTION - set an option
all - print options, current server and host
[no]debug - print debugging information
[no]d2 - print exhaustive debugging information
[no]defname - append domain name to each query
[no]recurse - ask for recursive answer to query
[no]search - use domain search list
[no]vc - always use a virtual circuit
domain=NAME - set default domain name to NAME
srchlist=N1[/N2/.../N6] - set domain to N1 and search list to N1,N2, etc.
root=NAME - set root server to NAME
retry=X - set number of retries to X
timeout=X - set initial time-out interval to X seconds
type=X - set query type (ex. A,ANY,CNAME,MX,NS,PTR,SOA,SRV)
querytype=X - same as type
class=X - set query class (ex. IN (Internet), ANY)
[no]msxfr - use MS fast zone transfer
ixfrver=X - current version to use in IXFR transfer request
server NAME - set default server to NAME, using current default server
lserver NAME - set default server to NAME, using initial server
finger [USER] - finger the optional NAME at the current default host
root - set current default server to the root
ls [opt] DOMAIN [> FILE] - list addresses in DOMAIN optional: output to FILE)
-a - list canonical names and aliases
-d - list all records
-t TYPE - list records of the given type (e.g. A,CNAME,MX,NS,PTR etc.)
view FILE - sort an ‘ls’ output file and view it with pg
exit - exit the program
12 Serv+ Ch_07 7/10/01 9:56 AM Page 377
C:\>nslookup www.newriders.com
Server: dns.myisp.com
Address: 10.18.6.103
Name: newriders.com
Address: 63.69.110.220
Aliases: www.newriders.com
ping
The Windows NT Server ping command is used to test communica-
tions between two network devices such as the current network
server and another server on the network. The flags for the
Windows NT Server ping command are as follows:
C:\>ping
Usage: ping [-t] [-a] [-n count] [-l size] [-f] [-i TTL] [-v TOS]
[-r count] [-s count] [[-j host-list] | [-k host-list]]
[-w timeout] destination-list
Options:
-t Ping the specified host until interrupted.
-a Resolve addresses to host names.
-n count Number of echo requests to send.
-l size Send buffer size.
-f Set Don’t Fragment flag in packet.
-i TTL Time To Live.
-v TOS Type Of Service.
-r count Record route for count hops.
-s count Time stamp for count hops.
-j host-list Loose source route along host-list.
-k host-list Strict source route along host-list.
-w timeout Timeout in milliseconds to wait for each reply.
12 Serv+ Ch_07 7/10/01 9:56 AM Page 378
C:\>ping 63.69.110.220
C:\>ping www.newriders.com
tracert
The Windows NT Server tracert command is similar to the ping
command, but it provides information about the routers that the
communications passes through going from the current network
server to another server on the network. The flags available for the
Windows NT Server tracert command are as follows:
C:\>tracert
Usage: tracert [-d] [-h maximum_hops] [-j host-list] [-w timeout] target_name
Options:
-d Do not resolve addresses to host names.
-h maximum_hops Maximum number of hops to search for target.
-j host-list Loose source route along host-list.
-w timeout Wait timeout milliseconds for each reply
12 Serv+ Ch_07 7/10/01 9:56 AM Page 379
C:\>tracert www.newriders.com
10 32 ms 31 ms 31 ms 287.ATM6-0.XR1.EWR1.ALTER.NET [152.63.20.253]
11 31 ms 31 ms 47 ms 193.ATM6-0.GW7.EWR1.ALTER.NET [152.63.24.205]
12 31 ms 31 ms 47 ms headland-media-gw.customer.ALTER.NET [157.130.19.94]
13 31 ms 31 ms 47 ms 63.69.110.220
Trace complete.
FIGURE 7.8
Windows 2000 Task Manager Performance tab.
FIGURE 7.9
Windows 2000 Task Manager Processes tab.
12 Serv+ Ch_07 7/10/01 9:56 AM Page 381
FIGURE 7.10
Windows 2000 Task Manager Applications tab.
Computer Management
The Windows 2000 Server System Information function is a replace-
ment for the Windows NT Server NT Diagnostics program. However,
the information is arranged very differently. Figure 7.11 shows an
example of the Windows 2000 Computer Management snap-in.
The Windows 2000 Computer Management snap-in contains
several nodes that can prove useful in providing information about
the configuration of the network server.
á System Summary node. This node displays general informa-
tion about the network server, including the version of
Windows 2000 installed, the system name, the system type,
the location of the Windows system directory, and statistics
about both physical and virtual memory.
á Hardware Resources node. This node displays all the hard-
ware settings in the network server, such as direct memory
access (DMA) channels, hardware interrupts (IRQs), I/O
addresses, and memory addresses of hardware components
installed in the network server.
á Components node. This node displays information about the
Windows 2000 configuration. It is useful in determining
information about device drivers and networking. This
includes the history of drivers installed on the network server.
12 Serv+ Ch_07 7/10/01 9:56 AM Page 382
FIGURE 7.11
Windows 2000 Computer Management snap-in.
Event Viewer
The Windows 2000 Server Event Viewer is similar in function to the
Windows NT Server Event Viewer, but is very different in appear-
ance. The appearance differs because the Windows 2000 Event
Viewer has been rewritten and it is now a snap-in to the Microsoft
Management Console (MMC).
You can view at least three logs with the Windows 2000 Event
Viewer: the System log, the Security log, and the Applications log.
In addition, you can view several other logs depending on the ser-
vices installed on the Windows 2000 Server. For example, if the
Domain Name System (DNS) service is installed on the Windows
2000 Server, there will be a DNS log listed in Event Viewer.
12 Serv+ Ch_07 7/10/01 9:56 AM Page 383
FIGURE 7.12
Windows 2000 Event Viewer System log.
FIGURE 7.13
Windows 2000 Event Viewer Security log.
FIGURE 7.14
Windows 2000 Event Viewer Applications log.
12 Serv+ Ch_07 7/10/01 9:56 AM Page 385
USAGE:
ipconfig [/? | /all | /release [adapter] | /renew [adapter]
| /flushdns | /registerdns
| /showclassid adapter
| /setclassid adapter [classidtoset] ]
C:\>ipconfig
IP Address. . . . . . . . . . . . : 192.168.1.130
Subnet Mask . . . . . . . . . . . : 255.255.255.0
Default Gateway . . . . . . . . . :
C:\>ipconfig /all
nbtstat
The Windows 2000 Server nbtstat command enables you to
display information about sessions that are using NetBIOS over
TCP/IP (NBT). The flags available for the Windows 2000 Server
nbtstat command are as follows:
C:\>nbtstat /?
Displays protocol statistics and current TCP/IP connections using NBT (NetBIOS over TCP/IP).
nbtstat [ [-a RemoteName] [-A IP address] [-c] [-n] [-r] [-R] [-RR] [-s] [-S] [interval] ]
-a (adapter status) Lists the remote machine’s name table given its name.
-A (Adapter status) Lists the remote machine’s name table given its IP address.
-c (cache) Lists NBT’s cache of remote [machine] names and their IP addresses.
-n (names) Lists local NetBIOS names.
-r (resolved) Lists names resolved by broadcast and via WINS.
-R (Reload) Purges and reloads the remote cache name table.
-S (Sessions) Lists sessions table with the destination IP addresses.
-s (sessions) Lists sessions table converting destination IP
➥addresses to computer NETBIOS names.
-RR (ReleaseRefresh) Sends Name Release packets to WINs and then starts Refresh.
netdiag
The Windows 2000 Server netdiag command runs a standard
set of network tests and generates a report of the results. The
Windows 2000 Server netdiag command is not part of a
standard Windows 2000 Server installation. It must be installed
from the Windows 2000 Server CD-ROM from the Support
Tools folder. The Windows 2000 netdiag command has one
very nice feature: It can be used without any flags at all and it
will perform a complete set of network tests. A server hardware
specialist just needs to analyze the output from the netdiag
command looking for the word Failed to find possible network
problems. Even though the netdiag command can be used without
any flags, several are available. The flags that can be used with the
Windows 2000 Server netdiag command are as follows:
C:\>netdiag /?
C:\>netdiag
.....................................
Global results:
Note that in the output of this netdiag command that the default
gateway test failed, because no default gateway address is defined on
this particular network server.
netstat
The Windows 2000 Server netstat command enables you to display
information about current TCP/IP connections to the network
server. The flags for the Windows 2000 Server netstat command
are as follows:
C:\>netstat /?
C:\>netstat -s
IP Statistics
Datagrams Forwarded = 0
Unknown Protocols Received = 0
Received Packets Discarded = 0
Received Packets Delivered = 2411236
Output Requests = 137224
Routing Discards = 0
Discarded Output Packets = 0
Output Packet No Route = 0
Reassembly Required = 0
Reassembly Successful = 0
Reassembly Failures = 0
Datagrams Successfully Fragmented = 0
Datagrams Failing Fragmentation = 0
Fragments Created = 0
ICMP Statistics
Received Sent
Messages 43 72
Errors 0 0
Destination Unreachable 5 34
Time Exceeded 0 0
Parameter Problems 0 0
Source Quenches 0 0
Redirects 0 0
Echos 38 0
Echo Replies 0 38
Timestamps 0 0
Timestamp Replies 0 0
Address Masks 0 0
Address Mask Replies 0 0
TCP Statistics
UDP Statistics
nslookup
The Windows 2000 Server nslookup command tests DNS service IP
name resolution. The Windows 2000 Server nslookup command can
be used in two different modes: command line and interactive. The
commands available in the interactive version of the Windows 2000
nslookup command are as follows:
> help
Commands: (identifiers are shown in uppercase, [] means optional)
NAME - print info about the host/domain NAME using default server
NAME1 NAME2 - as above, but use NAME2 as server
help or ? - print info on common commands
set OPTION - set an option
all - print options, current server and host
[no]debug - print debugging information
[no]d2 - print exhaustive debugging information
[no]defname - append domain name to each query
[no]recurse - ask for recursive answer to query
[no]search - use domain search list
[no]vc - always use a virtual circuit
domain=NAME - set default domain name to NAME
srchlist=N1[/N2/.../N6] - set domain to N1 and search list to N1,N2, etc.
root=NAME - set root server to NAME
retry=X - set number of retries to X
timeout=X - set initial time-out interval to X seconds
type=X - set query type (ex. A,ANY,CNAME,MX,NS,PTR,SOA,SRV)
querytype=X - same as type
class=X - set query class (ex. IN (Internet), ANY)
[no]msxfr - use MS fast zone transfer
ixfrver=X - current version to use in IXFR transfer request
server NAME - set default server to NAME, using current default server
lserver NAME - set default server to NAME, using initial server
finger [USER] - finger the optional NAME at the current default host
root - set current default server to the root
ls [opt] DOMAIN [> FILE] - list addresses in DOMAIN (optional: output to FILE)
-a - list canonical names and aliases
12 Serv+ Ch_07 7/10/01 9:57 AM Page 395
C:\>nslookup www.microsoft.com
Server: dns.myisp.net
Address: 192.168.6.103
Non-authoritative answer:
Name: www.microsoft.akadns.net
Addresses: 207.46.230.218, 207.46.131.91, 207.46.230.219, 207.46.131.199 207.46.230.229
Aliases: www.microsoft.com
Non-authoritative answer:
Name: newriders.com
Address: 63.69.110.220
ping
The Windows 2000 Server ping command enables you to test com-
munications between the current network server and another server
on the network. The flags for the Windows 2000 Server ping com-
mand are as follows:
C:\>ping /?
Usage: ping [-t] [-a] [-n count] [-l size] [-f] [-i TTL] [-v TOS]
[-r count] [-s count] [[-j host-list] |[-k host-list]]
[-w timeout] destination-list
Options:
-t Ping the specified host until stopped.
To see statistics and continue –type Control-Break;
➥To stop – type Control-C.
12 Serv+ Ch_07 7/10/01 9:57 AM Page 396
C:\>ping 192.168.1.200
C:\>ping london.nwtraders.msft
C:\>tracert
Usage: tracert [-d] [-h maximum_hops] [-j host-list] [-w timeout] target_name
Options:
-d Do not resolve addresses to host names.
-h maximum_hops Maximum number of hops to search for target.
-j host-list Loose source route along host-list.
-w timeout Wait timeout milliseconds for each reply.
C:\>tracert www.newriders.com
7 10 ms 10 ms 20 ms 153.at-1-0-0.XR1.ATL1.ALTER.NET [152.63.84.26]
8 10 ms 10 ms 20 ms 0.so-3-0-0.TR1.ATL1.ALTER.NET [152.63.10.69]
9 30 ms 30 ms 40 ms 109.at-5-0-0.TR1.NYC9.ALTER.NET [146.188.141.74]
12 Serv+ Ch_07 7/10/01 9:57 AM Page 398
10 30 ms 40 ms 30 ms 287.ATM6-0.XR1.EWR1.ALTER.NET [152.63.20.253]
11 31 ms 40 ms 30 ms 193.ATM7-0.GW7.EWR1.ALTER.NET [152.63.24.197]
12 30 ms 40 ms 40 ms headland-media-gw.customer.ALTER.NET [157.130.19.94]
13 30 ms 30 ms 40 ms 63.69.110.220
Trace complete.
pathping
The Windows 2000 Server pathping command is a combination of
the ping command and the tracert command. The flags that can be
used with the Windows 2000 pathping command are as follows:
C:\>pathping
Options:
-n Do not resolve addresses to host names.
-h maximum_hops Maximum number of hops to search for target.
-g host-list Loose source route along host-list.
-p period Wait period milliseconds between pings.
-q num_queries Number of queries per hop.
-w timeout Wait timeout milliseconds for each reply.
-T Test connectivity to each hop with Layer 2 priority tags.
-R Test whether each hop is RSVP aware.
C:\>pathping www.newriders.com
Tracing route to newriders.com [63.69.110.220]
over a maximum of 30 hops:
0 [172.16.34.132]
1 [172.16.34.254]
2 [172.16.33.1]
3 [172.16.250.26]
4 [172.16.250.9]
5 204.194.39.157
6 Serial5-1-0.GW1.JAX1.ALTER.NET [157.130.68.173]
7 153.at-1-0-0.XR1.ATL1.ALTER.NET [152.63.84.26]
8 0.so-3-0-0.TR1.ATL1.ALTER.NET [152.63.10.69]
12 Serv+ Ch_07 7/10/01 9:57 AM Page 399
9 109.at-5-0-0.TR1.NYC9.ALTER.NET [146.188.141.74]
10 287.ATM6-0.XR1.EWR1.ALTER.NET [152.63.20.253]
11 193.ATM7-0.GW7.EWR1.ALTER.NET [152.63.24.197]
12 headland-media-gw.customer.ALTER.NET [157.130.19.94]
13 .63.69.110.220
Trace complete.
Novell NetWare
Novell NetWare includes useful system and network diagnostic tools.
12 Serv+ Ch_07 7/10/01 9:57 AM Page 400
UNIX/Linux
The various versions of UNIX/Linux have their own (usually
unique) software tools that you can use for diagnostic purposes.
>ifconfig -a
lo0: flags=e08084b<UP,BROADCAST,LOOPBACK,RUNNING,SIMPLEX,MULTICAST,GROUPRT,64BIT
>
inet 127.0.0.1 netmask 0xff000000 broadcast 127.255.255.255
inet6 ::1/0
fi0: flags=e0a0843<UP,BROADCAST,RUNNING,SIMPLEX,ALLCAST,MULTICAST,GROUPRT,64BIT>
inet 192.168.197.3 netmask 0xffffff00 broadcast 192.168.197.255
en1: flags=e080863<UP,BROADCAST,NOTRAILERS,RUNNING,SIMPLEX,MULTICAST,GROUPRT,64BIT>
inet 192.168.2.23 netmask 0xffffff00 broadcast 192.168.2.255
netstat
The netstat utility is used to display TCP/IP network statistics. For
specifics on the netstat command, enter the following command at
the UNIX/Linux prompt: man netstat. The information displayed
by the man command also tells how to interpret the output generated
by the command.
The following listing is an example of UNIX/Linux netstat output:
>netstat -s
ip:
1071952632 total packets received
0 bad header checksums
0 with size smaller than minimum
0 with data size < data length
12 Serv+ Ch_07 7/10/01 9:57 AM Page 403
echo: 475341
#10: 2
time exceeded: 10835
parameter problem: 2
475341 message responses generated
igmp:
128264 messages received
0 messages received with too few bytes
0 messages received with bad checksum
128247 membership queries received
0 membership queries received with invalid field(s)
17 membership reports received
0 membership reports received with invalid field(s)
17 membership reports received for groups to which we belong
3 membership reports sent
tcp:
969995197 packets sent
829617186 data packets (3708222803 bytes)
3802849 data packets (1840909853 bytes) retransmitted
42344148 ack-only packets (31663214 delayed)
388 URG only packets
43394 window probe packets
89208904 window update packets
4978328 control packets
1056319126 packets received
365380856 acks (for 3429378578 bytes)
9787520 duplicate acks
5 acks for unsent data
877867036 packets (3622795012 bytes) received in-sequence
1108797 completely duplicate packets (184293621 bytes)
199773 packets with some dup. data (4072561 bytes duped)
5051682 out-of-order packets (751752818 bytes)
2507 packets (331001 bytes) of data after window
1514 window probes
2358506 window update packets
184753 packets received after close
0 packets with bad hardware assisted checksum
13521 discarded for bad checksums
48 discarded for bad header offset fields
1487 connection request
549677 connection requests
4595174 connection accepts
4933946 connections established (including accepts)
6188958 connections closed (including 682010 drops)
1190931 embryonic connections dropped
12 Serv+ Ch_07 7/10/01 9:57 AM Page 405
ping
The ping command enables you to test TCP/IP communications
between two devices on the network, such as two network servers.
For specifics on the ping command, enter the following command at
the UNIX/Linux prompt: man ping. The information displayed by
the man command also tells how to interpret the output generated by
the command.
The following listing is an example of UNIX/Linux ping output:
>ping -c 10 www.newriders.com
PING newriders.com: (63.69.110.220): 56 data bytes
64 bytes from 63.69.110.220: icmp_seq=0 ttl=244 time=38 ms
64 bytes from 63.69.110.220: icmp_seq=1 ttl=244 time=37 ms
64 bytes from 63.69.110.220: icmp_seq=2 ttl=244 time=37 ms
64 bytes from 63.69.110.220: icmp_seq=3 ttl=244 time=38 ms
12 Serv+ Ch_07 7/10/01 9:57 AM Page 406
traceroute
The traceroute command is also used to test TCP/IP communica-
tions between two network servers, but it also reports the routers
that the communications packets travel through going from one net-
work device, such as a network server, to another network device.
For specifics on the traceroute command, enter the following com-
mand at the UNIX/Linux prompt: man traceroute. The informa-
tion displayed by the man command also tells how to interpret the
output generated by the command.
The following listing is an example of UNIX/Linux traceroute
output:
>traceroute www.newriders.com
trying to get source for www.newriders.com
source should be 192.168.2.23
traceroute to newriders.com (63.69.110.220) from 192.168.2.23, 30
hops max
outgoing MTU = 1500
1 vlan602.rsm.bfs.fsu.edu (192.168.2.1) 4 ms 1 ms 1 ms
2 fe600.c7507.bfs.fsu.edu (192.168.250.18) 1 ms 1 ms 2 ms
3 atm5001.c7507.dsl.fsu.edu (192.168.250.54) 2 ms 2 ms 2 ms
4 204.194.39.157 (204.194.39.157) 4 ms 4 ms 4 ms
5 Serial5-1-0.GW1.JAX1.ALTER.NET (157.130.68.173) 7 ms 7 ms 7 ms
6 153.at-2-1-0.XR1.ATL1.ALTER.NET (152.63.84.34) 16 ms * *
7 0.so-3-0-0.TR1.ATL1.ALTER.NET (152.63.10.69) 15 ms 16 ms 15 ms
8 109.at-5-0-0.TR1.NYC9.ALTER.NET (146.188.141.74) 36 ms 35 ms 36 ms
9 287.ATM6-0.XR1.EWR1.ALTER.NET (152.63.20.253) 38 ms 38 ms 39 ms
10 193.ATM6-0.GW7.EWR1.ALTER.NET (152.63.24.205) 37 ms 37 ms 37 ms
11 headland-media-gw.customer.ALTER.NET (157.130.19.94) 39 ms 39 ms 39 ms
12 web05.ot.hdmss.net (63.69.110.67) 56 ms * 38 ms
12 Serv+ Ch_07 7/10/01 9:57 AM Page 407
tcpcfg
The tcpcfg command will display the current TCP/IP configuration
for OS/2 Warp Server. Once issued it brings up the same window as
shown in Figure 7.15.
ping
The ping command in OS/2 Warp Server is similar in function to
the ping command in other network server operating systems. The
format of the OS/2 ping command is as follows:
ping ipaddress
12 Serv+ Ch_07 7/10/01 9:57 AM Page 408
FIGURE 7.15
OS/2 TCP/IP configuration.
netstat
The OS/2 netstat command is similar to the netstat command
in Microsoft’s Windows NT Server, Windows 2000 Server, and
UNIX/Linux systems. The flags for the OS/2 netstat command
are as follows:
netstat [-a] [-r] [-s] [-n] [-p] [-t] [-u] [-i] [-c] [-m]
ifconfig
The OS/2 ifconfig command is similar to the ifconfig command
found in UNIX/Linux systems. It displays the TCP/IP configuration
information for a particular NIC.
12 Serv+ Ch_07 7/10/01 9:57 AM Page 409
ifconfig interface
Where interface can be any of the following:
Interface Description
lo loopback
lan0 thru lan7 LAN interfaces (Ethernet, Token Ring)
[C:tcpipbin]ifconfig lan0
lan0: flags=3030<UP,BROADCAST,NOTRAILERS,RUNNING,BRIDGE,SNAP>
inet 192.168.1.202 netmask ffffff00x broadcast 192.168.1.255
tracerte
The OS/2 tracerte command is similar in function to the Windows
NT Server and the Windows 2000 Server Tracert command and the
UNIX/Linux traceroute command. It provides a list of routers that
a packet travels through as it goes from the network server that
issued the tracerte command to the destination network device.
The format of the OS/2 tracerte command is as follows:
tracerte ipaddress
System Shutdown
To perform some troubleshooting activities, you must shut down the
network server. The different network server operating systems have
different procedures or commands to shut down the system. That
said, some general steps should always be followed when shutting
down the network server.
Always inform the users that you are going to shut down the system,
by broadcasting a message to all users letting them know that the
network server is going to be shut down. Allow enough time for the
users to save their work and to log off of the network server.
Typically a message such as
12 Serv+ Ch_07 7/10/01 9:57 AM Page 410
down the different network server To shut down a Microsoft Windows NT network server, follow
operating systems. these steps:
EXAM
or
1. Open a command prompt and use the NET SEND command.
2. Select Start, Programs, Command Prompt.
3. In the command-prompt window, type the net send
“message” command.
or
1. Press Ctrl+Alt+Del, click the Shut Down button, click Shut
Down, and then click OK.
or
1. Click Start, Programs, Accessories, Command Prompt.
2. In the command-prompt window, type the net send
“message” command.
or
1. Press Ctrl+Alt+Del, and then select Shut Down from the
drop-down box. Click OK.
UNIX Shutdown
Remember that all commands and their flags in UNIX are case sen-
sitive. All commands are always entered using lowercase characters.
To send a message to all users who are logged in to the UNIX net-
work server, use the UNIX wall command, as follows:
wall message
Linux Shutdown
Like other versions of UNIX, Linux includes a wall command that
can be used to notify all logged-in users that the network server is
going to be shutdown. However, the Linux wall command uses a
file to contain the message. The format for the Linux wall com-
mand is as follows:
wall [file]
Because in both UNIX and Linux a file can be standard input, how-
ever, the message for the wall command could be created this way:
wall
ctrl+d
The time parameter is required. The word now may be used to indicate
immediate shutdown. Here are several examples of Linux shutdowns:
Example 1:
/sbin/shutdown –r 17:00 server shutting down at 5:00pm
The preceding command halts (shuts down) the server at 5 p.m., and
will send the message “server shutting down at 5:00pm” to all the users.
Example 3:
/sbin/shutdown –c
To shut down IBM OS/2 use the following command from a com-
mand prompt:
shutdown
á CD-ROM drive
á DVD-ROM drive
á Floppy disk drive
á Hard disk drive
á System Board
á SCSI bus controller
á Video controller
á Network interface card
á Processor
á Cooling fans
á Memory
Using Documentation
The network server logbook can be a most useful tool in trouble-
shooting. Assuming that the logbook has been kept up-to-date,
reading the last few pages of the log can provide a valuable clue
that can aid in troubleshooting. Perhaps a new software package
has recently been installed on the network server. Or maybe a new
hardware component was recently added to the network server.
The logbook proves it value when it holds a clue that helps solve
a problem with the network server.
and Novell) offer through an annual subscription service a set of Add to Tool Kit A server hardware
technical support CDs that are updated on a monthly basis. These specialist’s tool kit should include a
CDs contain not only recent patches to the operating system, but technical support subscription service
they also contain many articles or white papers with technical details for the network operating system
most often en-countered by the server
of the operating system, knowledge bases of reported problems and
hardware specialist.
their solutions, resource kits, and utilities.
Microsoft’s technical support subscription service is called TechNet.
Novell calls their technical support subscription service the Novell
Support Connection CD. Both of these services are extremely valuable
to the server hardware specialist.
12 Serv+ Ch_07 7/10/01 9:57 AM Page 418
Newsgroups
Newsgroups are accessed by using a newsreader such as Microsoft
Outlook Express or the newsreader option in NetScape. The follow-
ing sections list some of the more useful newsgroups for the major
operating systems.
comp.os.ms-windows.nt.admin.misc
comp.os.ms-windows.nt.admin.networking
comp.os.ms-windows.nt.admin.security
comp.os.ms-windows.nt.misc
comp.os.ms-windows.nt.setup.hardware
comp.os.ms-windows.nt.setup.misc
comp.os.ms-windows.nt.software.services
comp.os.os2.comm
comp.os.os2.misc
comp.os.os2.networking.misc
comp.os.os2.setup.misc
comp.os.os2.utilities
comp.os2.networking.tcp-ip
12 Serv+ Ch_07 7/10/01 9:57 AM Page 419
comp.os.netware.misc
comp.os.netware.security
UNIX Newsgroups
comp.unix.admin
comp.unix.misc
comp.unix.pc-clone.16bit
comp.unix.pc-clone.32bit
comp.unix.questions
comp.unix.sco.misc
comp.unix.solaris
comp.unix.unixware.misc
Linux Newsgroups
comp.os.linux.answers
comp.os.linux.hardware
comp.os.linux.misc
comp.os.linux.networking
comp.os.linux.security
comp.os.linux.setup
Server Vendors
Compaq www.compaq.com
Dell www.dell.com
Hewlett-Packard www.hp.com
IBM www.ibm.com
Unisys www.unisys.com
12 Serv+ Ch_07 7/10/01 9:57 AM Page 420
Novell www.novell.com
UNIX www.sco.com
www.sun.com
Linux www.redhat.com
www.caldera.com
www.suse.com
IBM www.ibm.com
Gathering Resources
When a problem has been reported with the network server, it is
time to shift into “troubleshooting mode.” Many things need to be
done to effectively troubleshoot the problem. The first step is to
gather all the information about the problem and the resources that
will be needed to help troubleshoot the problem. The resources to
be gathered depend on the apparent source of the problem. A net-
work performance problem will probably need a network analysis
tool. An apparent disk failure will require the ability to check
whether the disk drive has actually failed (using senses to check for
loose cables, checking for lack of power, checking for error LEDs on
the disk drive, touching the failed drive to feel whether it is spin-
ning, and so on). That spare external SCSI cable might come in
handy in this situation. Failure of the network server to power up
will probably require the use of a POST Diagnostics Board. There is
always the possibility that you might not be able to solve the prob-
lem with the network server. If the problem is beyond your ability to
solve, call for help from a senior server hardware specialist. Or if you
suspect that the problem is actually a network problem and not a
problem with the network server or its NIC, it is time to call in the
network specialist. Remember the goal is get the network server back
in operation as quickly as possible. If that means making a few
phone calls, make the phone calls.
As a standard part of the troubleshooting process, the server hard-
ware specialist should make sure that he or she has all the available
documentation, the network server logbook, errors logs, event logs,
answers to the questioning process, and support CDs. If you suspect
12 Serv+ Ch_07 7/10/01 9:57 AM Page 421
IDENTIFYING BOTTLENECKS
. 6.3 Identify bottlenecks (e.g., processor, bus transfer, I/O,
disk I/O, network I/O, memory)
To be able to diagnose which network server component is causing
a system bottleneck, the server hardware specialist must have access
to performance-monitoring software tools specific to the network
server operating system. Most network server operating systems
have at least a basic performance-monitoring software package
included with the operating system. Third-party vendors also have
performance-monitoring software tools available for most network
server operating systems.
Examples of performance-monitoring software tools for various net-
work server operating systems include the following:
á Performance Monitor. For Microsoft Windows NT Server
á Performance. For Microsoft Windows 2000 Server
á ManageWise. For Novell NetWare
á sar, iostat, vmstat, ps. For UNIX/Linux
á System Performance Monitor/2 (SPM/2). For IBM OS/2
FIGURE 7.16
Windows NT Server Performance Monitor.
FIGURE 7.17
Windows 2000 Server Performance.
>sar
3 3 4 000078638900 04/23/01
00:20:20 68 32 0 0
00:20:30 51 49 0 0
00:20:40 78 22 0 0
00:20:50 83 16 0 0
00:30:01 76 24 0 0
00:30:11 86 14 0 0
00:30:21 76 24 0 0
00:30:31 75 25 0 0
00:30:41 87 13 0 0
00:30:51 79 21 0 0
00:40:01 70 29 0 0
00:40:11 57 43 0 0
00:40:21 67 33 0 0
00:40:32 46 54 0 0
00:40:42 63 37 0 0
00:40:52 76 24 0 0
00:50:01 66 34 0 0
00:50:11 80 20 0 0
00:50:21 76 24 0 0
Average 78 22 0 0
>vmstat
kthr memory page faults cpu
——- —————- ———————————— —————— —————-
r b avm fre re pi po fr sr cy in sy cs us sy id wa
1 0 31888 144430 0 0 0 2 9 0 344 197 312 27 23 46 4
>iostat
The UNIX/Linux Processes (ps) command lists all the processes that
are running on the UNIX/Linux network server.
Output from a typical UNIX/Linux ps command looks like this:
>ps -A
PID TTY TIME CMD
0 - 46:02 swapper
1 - 118:31 init
516 - 64493:11 kproc
774 - 2:58 kproc
1032 - 794:35 kproc
1860 - 0:00 kproc
2234 - 0:09 errdemon
2364 - 558:09 syncd
2648 - 0:00 srcmstr
3396 - 33:04 syslogd
3642 - 0:00 lpd
3904 - 0:06 xntpd
4180 pts/12 0:23 pine
4388 - 0:00 portmap
4644 - 78:41 inetd
4918 - 0:06 kproc
5160 - 3395:21 kproc
8002 - 0:00 biod
8260 - 1:01 sshd2
9038 - 0:00 rpc.statd
9806 - 0:27 rpc.lockd
10776 pts/11 0:00 ksh
11172 - 0:00 sleep
11954 - 0:00 telnetd
12 Serv+ Ch_07 7/10/01 9:57 AM Page 427
Using the performance-monitoring tools available for the network network server.
server operating system, the server hardware specialist should be able
to monitor the utilization of the processor(s) in the network server.
The two counters to watch most closely are the percent processor
12 Serv+ Ch_07 7/10/01 9:57 AM Page 428
Know the options for eliminating a If the processor utilization percentage is constantly greater than
processor bottleneck. 70%, the processor is the bottleneck in the network server.
EXAM
The processor queue length is the number of processes waiting for the
processor. If the processor queue length is constantly greater than 2,
the processor is the bottleneck in the network server.
You can alleviate the processor bottleneck by doing any of the
following:
á Upgrade the existing processor to a faster processor.
á Add an additional processor to the network server.
á Move applications or services from this network server to
another network server.
Memory Bottlenecks
On network servers with operating systems that do not support virtual
memory (for instance, Novell NetWare 4 and below), memory bottle-
necks are rather easy to detect. Memory shortages manifest themselves
on these systems as the inability to load additional software or the lack
of “buffers” (temporary storage between the system and its disk dri-
ves), which are easily detected using standard server management soft-
ware. The solution to a memory shortage is to add more memory.
On network servers with operating systems that do support virtual
memory, several indicators or counters should be monitored. The
first counter is the amount of available memory (memory that is not
being used) after all the network server software has been loaded. To
totally prevent paging or swapping of memory pages to disk, just
have more actual or physical memory than is needed by all the soft-
ware running on the network server.
The other counter that should be monitored to check for memory
bottlenecks is the number of pages that are read from or written to the
paging file on the disk in 1 second. The number of pages per second
that is acceptable depends on the speed of the processor. Generally the
number should be less than 2,000 per second. Heavy paging often
manifests itself as high disk utilization, because the memory pages are
written to disk when the memory is needed by another program.
12 Serv+ Ch_07 7/10/01 9:57 AM Page 429
TIP
when there is an attempt to access a program page and the page is Know which statistics indicate that
memory is the bottleneck in a net-
no longer in memory, but it must be retrieved from disk. A page
EXAM
work server.
fault problem is indicated if the number of page faults per second is
more than double the number of page faults recorded on the server
baseline measurements taken under a normal workload.
To alleviate a memory bottleneck, add more memory to the net-
work server.
Eliminate Memory Bottleneck
TIP
Paging performance can often be improved by placing the paging file
(or the swap file) on the fastest disk in the network server and on a disk Know the options for eliminating a
memory bottleneck.
that doesn’t contain the network server operating system or the data
EXAM
files used by the major application running on the network server.
á Defragment the disk drive. Having fragmented files on the Bottleneck Know which statistics
disk drive can drastically reduce disk performance. indicate that the disk subsystem is
EXAM
neck. á Replace the disk controller with a disk controller that has a
built-in memory cache.
á Replace the disk drives with disk drives with a higher rota-
tional speed (and therefore faster access time).
á Replace the entire disk subsystem (controller and disk drives)
with a faster disk subsystem.
Indicators of a Network
Network Subsystem Bottlenecks
TIP
Subsystem Bottleneck Know Network subsystem bottlenecks are perhaps the most difficult to
which statistics indicate that the detect and solve. Typical performance-monitoring software can sup-
EXAM
network subsystem is the bottle- ply counts of packets sent and received and little else. To analyze
neck. network performance, a network analyzer (hardware or software) is
needed. A network analyzer can actually capture packets on the net-
work for later analysis. The percent network utilization, a very valu-
able indicator of network performance, can be determined easily. A
network analyzer can also determine the type of network packets (for
instance, broadcasts, multicasts, or unicasts) and the protocols that
are consuming the most network bandwidth.
A network utilization percentage that is constantly greater than
65% is generally considered to be high utilization, and can consti-
tute a bottleneck.
Solutions to a network subsystem bottleneck are generally expensive
Eliminating a Network Subsystem and include the following:
TIP
NOTE
trying to get to the network server, consider using adapter Two Versions of Microsoft Network
teaming to increase the network bandwidth available to the Monitor There are two different ver-
sions of Microsoft’s Network Monitor
network server. This requires multiple NICs to be installed
software. A cut-down version is
in the network server and a Fast Ethernet switch that supports
included with Microsoft Windows NT
bandwidth aggregation to be installed. (See the section
Server and Microsoft Windows 2000
“Adapter Teaming” in Chapter 1, “0.0—Planning and Server operating systems. This ver-
Specifications,” for more details.) sion can monitor only network traffic
á Eliminate unnecessary protocols from the network. When leaving or entering the network server
on which it is running. The full version
multiple protocols are installed on a network server, announce-
of Network Monitor is a component of
ments sent from the network server are sent once for each
the Microsoft System Management
installed protocol (thus increasing network traffic).
Server 2.0 BackOffice Product. The
The following are a couple of well-known network monitoring and full version of Network Monitor can
analysis tools: capture and analyze all packets on
the network.
á Microsoft Network Monitor
á Novell LANalyzer for Windows
NOTE
Add to Tool Kit A server hardware
Here is the short list of things to check if you suspect a disk drive
failure in the network server:
á Check to make sure that the data cable is firmly attached to
the disk drive.
á Check to make sure the disk drive is getting power. Check to
make sure that the power plug is plugged into the disk drive.
á By touching a disk drive, an experienced server hardware spe-
NOTE
Add to Tool Kit A server hardware cialist can actually feel the disk drive spinning. In a non-noisy
specialist’s tool kit should include environment, an experienced server hardware specialist can
a stethoscope for listening to hard also hear the disk drive spinning. However, very few network
disk drives. Can a Server Hardware servers are located in non-noisy environments. By using a
Specialist who uses a stethoscope to
stethoscope, a server hardware specialist can easily hear
listen to hard disks be referred to as
whether a hard disk drive is functioning normally.
a “spin doctor”?
á Any “clicking” noise coming from a hard disk drive is defi-
nitely a bad sign. It usually indicates a catastrophic failure of
the disk drive.
NOTE
does not exceed the maximum allowed for the SCSI signaling Add to Tool Kit A server hardware
system (single-ended, high-voltage differential, or low-voltage specialist’s tool kit should include
differential) being used. a known good external SCSI cable
with 68-pin HD connectors, a known
á Remember that to achieve their high data-transfer speeds, good external SCSI cable with 68-pin
some high-speed SCSI systems (such as Ultra 160) do not VHDCI connectors, and a known
allow the maximum number of devices to be attached to the good external SCSI cable with 50-pin
SCSI bus. (Usually the maximum is seven.) Some high-speed connectors.
SCSI systems limit the number of disk drives to three.
á Make sure that the SCSI cables are top quality. Quality cables
are needed to support high data-transfer speeds.
single ATA channel (ribbon cable) that one of the disk drives is Add to Tool Kit A server hardware
set to the master role and that the other is set to the slave role. specialist’s tool kit should include a
Alternatively, both disk drives may be set to cable select (CSEL). known good IDE/ATA cable and extra
IDE/ATA jumpers.
á Make sure that the ATA cable does not exceed the maximum
ATA cable length (usually 18 inches).
12 Serv+ Ch_07 7/10/01 9:57 AM Page 434
Add to Tool Kit A server hardware require an ATA cable that differs from regular ATA system cables.
specialist’s tool kit should include a The Ultra ATA systems require an 80-conductor cable with a 40-pin
known good Ultra ATA 80-conductor
connector rather than the 40-conductor cable with a 40-pin connec-
cable.
tor used by regular ATA system.
Add to Tool Kit A server hardware (known as 10/100 Ethernet NICs) generally have an LED to indicate
specialist’s tool kit should include a the speed at which the NIC is communicating.
Category 5 UTP path cable that is at
least 10 feet long and is known to be If the network cable is attached to both the NIC and a hub or
in good working condition. switch and there is still a problem with the network communication,
replace the network patch cable with a cable that is known to work.
If the network still does not communicate, the problem is either
with the NIC (which should be replaced with a known working
NIC) or with the port on the hub/switch. Try using a different port
on the hub/switch.
If none of these changes clear up the network problem, it is time to
call the network expert on your contact list.
C:\>ping 192.168.1.200
C:\>ping london.nwtraders.msft
C:\>ping seattle.namerica1.nwtraders.msft
Unknown host seattle.namerica1.nwtraders.msft.
C:\>ping 10.2.3.12
C:\>ping vancouver.namerica1.nwtraders.msft
The Request timed out message indicates that the pinged computer
NOTE
Add to Tool Kit A server hardware did not reply to the ping command. This usually means that the
specialist’s tool kit should include a computer being pinged is not currently running. On the plus side,
POST Diagnostics Board designed for the name vancouver.namerica1.nwtraders.msft was resolved to the
a PCI bus.
address 192.168.1.112 as shown in the output of the ping com-
mand. Also because the message is Request timed out rather than
Destination host unreachable, the default gateway was properly
configured on the computer issuing the ping command.
NOTE
Diagnostics Board.
the purpose of a POST Diagnostics
Board. A POST Diagnostics Board is a diagnostic device that plugs into any
EXAM
available PCI expansion slot in the network server and tests the
operation of the system as it powers up. Any error condition displays
on a set of LEDs or LCDs mounted on the POST Diagnostics
Board. The meaning of the error code and the action to take to
resolve the boot failure are included with the documentation of the
POST Diagnostics Board.
CHAPTER SUMMARY
This chapter covered troubleshooting. You learned about the uses of
KEY TERMS
questioning techniques and the use of your senses as aids in trouble-
shooting network server problems. The value of knowing whom to • APIPA
contact for specific problems and failures was discussed. The names • Event Viewer
and examples of common troubleshooting and diagnostic tools for • FRU
various network server operating systems were presented. The proper • hostname
commands used to shut down various network server operating sys-
tems were also documented. • ifconfig
• iostat
The ways to determine network server bottlenecks were presented
and solutions to the various bottlenecks were discussed. Various • ipconfig
troubleshooting techniques for different hardware components were • Magic packet
explained. • man
• nbtstat
• netstat
• nslookup
• pathping
• ping
• POST Diagnostics Board
• ps
• sar
• shutdown
• Task Manager
• traceroute
• tracert
• tracerte
• vmstat
• Wake-On-LAN
• wall
• Windows NT Diagnostics
12 Serv+ Ch_07 7/10/01 9:57 AM Page 440
A P P LY Y O U R K N O W L E D G E
Exercises 3. Ping the IP address of the default gateway and
note the results.
7.1 Using Network Tools and Utilities 4. Ping the IP address of any remote computer (a
It’s time to try some of these network utilities. computer on the other side of the router) and
Windows 2000 Server is used as the example in this note the results.
exercise, but you can use any network server operating 5. Perform a Tracert to the same IP address that you
system, although some of the actual commands will used in step 4 and note the results.
have to be changed (for example, Windows 2000 server
uses ipconfig to obtain TCP/IP configuration informa-
tion and UNIX/Linux uses ifconfig). 7.3 Using Other TCP/IP Utilities
Estimated Time: 15 minutes. Now try a few of the other TCP/IP utilities to see how
they work.
Use the network troubleshooting tools you have avail-
able to answer the following questions about the Estimated Time: 20 minutes.
TCP/IP configuration: 1. Try the netstat command to get TCP/IP statis-
What is the host name of the network server? (Hint: tics. The Windows 2000 netstat command to do
Use the hostname command.) this is netstat –s.
What is the TCP/IP address of the network server? 2. Try the nslookup command to look up the IP
(Hint: Use ipconfig /all.) address of the following IP names:
www.microsoft.com
What is the TCP/IP subnet mask?
www.newriders.com
What is the IP address of default gateway?
www.comptia.org
What is the IP address of the DNS server?
The format of the nslookup command to do
7.2 Testing the Network these is:
Now that you have gathered information about the nslookup ipname
TCP/IP configuration, use this information to test
the network.
Estimated Time: 15 minutes. Review Questions
1. Ping the loopback address (127.0.0.1) and note 1. You’ve just arrived at a client site and when you
the results. enter the server room you notice that it seems to
2. Ping the IP address of the network server and be rather warm. Who should you contact to dis-
note the results. cuss the temperature of the server room?
12 Serv+ Ch_07 7/10/01 9:57 AM Page 441
A P P LY Y O U R K N O W L E D G E
2. What processor utilization percentage indicates 15. What is the name of the utility in Microsoft
that the processor is the bottleneck in the net- Windows NT Server and Microsoft Windows
work server? 2000 that will display the processes running on
the network server?
3. When analyzing network server bottlenecks, a
lack of memory when you are using an operating
system that supports virtual memory may mani-
fest itself as what? Exam Questions
4. How are memory bottlenecks eliminated? 1. A network interface card’s Wake-On-LAN tech-
5. What disk subsystem utilization percentage indi- nology is activated when the NIC receives a spe-
cates that the disk subsystem is the bottleneck in cial TCP/IP packet. What is the special TCP/IP
the network server? packet called?
6. What piece of equipment is used to troubleshoot A. A power-up packet
a network server boot failure? B. A magic packet
7. What is the purpose of Wake-On-LAN C. A cookie packet
technology?
D. A startup packet
8. What is the Windows NT Server command-line
utility that will display the current TCP/IP con- 2. You have just run a performance test on a net-
figuration? work server and obtained the following results:
11. What is the name of the TCP/IP utility used to Disk Subsystem: Disk Queue Length 1
test name resolution from a DNS server? Network: Percent Network Utilization 20%
12. What is the UNIX/Linux command that will Based on this information, what is the network
send a message to all users logged in to the net- server bottleneck?
work server?
A. Processor
13. What is the proper way to shut down a Windows
NT/Windows 2000 network server? B. Memory
A P P LY Y O U R K N O W L E D G E
3. Which of the following commands is used to 6. Under what conditions should you consider the
shut down a Novell NetWare network server? use of a POST Diagnostics Board?
A. shutdown A. When the network server is performing
poorly
B. down
B. When the network server will not boot
C. stop
C. When the network server will not connect to
D. exit
the network
4. Which indicators (statistics) should be monitored
D. When the network server services file to load
to determine whether the processor is the bottle-
neck in the network server? 7. You have just run a performance test on a net-
work server and obtained the following results:
i. Percent processor utilization
Processor: Percent Utilization 30%
ii. Percent disk utilization
Processor: Processor Queue Length 1
iii. Processor queue length
Memory: Pages per second 15020
iv. Disk queue length
Disk Subsystem: Percent Disk
A. i and ii
Utilization 99%
B. i, ii, and iv
Disk Subsystem: Disk Queue Length 5
C. i and iii
Network: Percent Network Utilization 15%
D. iii and iv
Based on this information, what is the network
5. Which TCP/IP utility is used to test network server bottleneck?
communications from the network server to
A. Processor
another server on the network?
B. Memory
A. IPCONFIG
C. Disk subsystem
B. NETSTAT
D. Network subsystem
C. Ping
D. NETDIAG
12 Serv+ Ch_07 7/10/01 9:57 AM Page 443
A P P LY Y O U R K N O W L E D G E
8. Which of the following are acceptable ways to 10. Which indicators (statistics) should be monitored
eliminate processor bottlenecks? to help determine whether the disk subsystem is
the bottleneck in the network server?
i. Upgrade the existing processor to a faster
processor. i. Percent disk utilization
ii. Add another processor to the network server. ii. Percent processor utilization
iii. Move applications or services to another iii. Disk queue length
network server.
iv. Processor queue length
iv. Add more memory to the network server.
A. i, ii, iii, and iv
A. i, ii, iii, and iv
B. i and ii
B. i and iv
C. iii and iv
C. i, ii, and iii
D. i and iii
D. i only
11. What is the TCP/IP utility that will report the
9. Which of the following methods can be used routers that a communications packet passes
to eliminate a memory bottleneck on a net- through when traveling from one network server
work server? to another network server?
i. Install faster memory A. netstat
ii. Add more memory B. ping
iii. Install a slower processor C. tracert
iv. Replace SIMMs with DIMMs D. route
A. i and iii 12. What is the very first step that should be taken
as part of the process of shutting down a net-
B. ii only
work server?
C. iv only
A. Notify logged-in users that the network server
D. i only will be shut down shortly.
B. Turn the network server power switch off.
C. Issue the shutdown command (or equivalent)
for the network server operating system.
D. Press Ctrl+Alt+Del and stand back.
12 Serv+ Ch_07 7/10/01 9:57 AM Page 444
A P P LY Y O U R K N O W L E D G E
13. A new hard disk drive has been added to a SCSI A. The DNS server could not resolve the name of
channel. However, when the network server boots the host computer being pinged.
the SCSI controller not only doesn’t it see the new
B. There is no default gateway configured on the
disk drive, but one of the disk drives that it previ-
network server that issued the Ping command.
ously saw no longer appears in the list of devices
attached to the SCSI controller. Which of the fol- C. The pinged computer did not reply to the ping.
lowing is the most likely cause of the problem? D. The Ping service is not running on the destina-
A. The new SCSI disk drive has the same SCSI tion computer.
ID as the SCSI device already attached to the
SCSI channel.
B. The SCSI bus is not terminated at both ends. Answers to Review Questions
C. The SCSI channel is too long. 1. You should contact the person responsible for
D. The SCSI disk drive is set to master rather HVAC for the building where the server room is
than slave. located about the temperature in the server room.
For more information, see the section “Identify
14. A second disk drive has been installed on an ATA Contacts.”
disk channel that already had a disk drive attached
to it. However, the network server now fails to 2. If the processor utilization percentage is con-
boot. What is the most likely cause of the problem? stantly over 70%, the processor is the bottleneck
in the network server. For more information, see
A. The ATA disk channel is not terminated the section “Processor Bottlenecks.”
correctly.
3. The lack of memory in a network operating
B. The ATA disk channel is too long. system that supports virtual memory causes much
C. The ATA disk drives are misconfigured; one paging activity. This increases disk subsystem uti-
disk drive must be set to the master role, and lization. Even though it may look like the disk
the other must be set to the slave role. subsystem is too slow, the high utilization of the
disk is caused by lack of memory. For more infor-
D. The ATA disk drive ID number of the new mation, see the section “Memory Bottlenecks.”
disk drive conflicts with the ATA disk drive
ID number of the existing disk drive. 4. A memory bottleneck is eliminated by adding more
memory to the network server. For more informa-
15. You issue the following Ping command: tion, see the section “Memory Bottlenecks.”
ping 10.20.30.40
5. If the disk subsystem utilization is constantly over
And you receive the following message: 90%, the disk subsystem is the bottleneck in the
network server. For more information, see the
Destination host unreachable
section “Disk Subsystem Bottlenecks.”
What does this message indicate?
12 Serv+ Ch_07 7/10/01 9:57 AM Page 445
A P P LY Y O U R K N O W L E D G E
6. A POST Diagnostics Board is used to help 12. The UNIX/Linux command that will send a
troubleshoot network server boot problems. message to all users logged in to the network
For more information, see the section server is the wall (write all) command. For more
“Troubleshooting Boot Failure.” information, see the sections “UNIX Shutdown”
or “Linux Shutdown.”
7. Wake-On-LAN (WOL) technology allows a net-
work interface card to power up the computer in 13. A Windows NT/Windows 2000 network server
which it is installed upon receiving a special can be shut down in several ways. Use the Start
packet (known as the WOL Magic packet). For menu, select Shut Down, and then select Shut
more information, see the section “Remote Down the Server. Press Ctrl+Alt+Del, click the
Troubleshooting for a Wake-On-LAN.” Shutdown button, click Shutdown, and then
click OK or Yes. For more information, see the
8. The Windows NT Server command-line utility
sections “Windows NT Server Shutdown” or
that will display the current TCP/IP configura-
“Windows 2000 Server Shutdown.”
tion is ipconfig. For more information, see the
section “Network Diagnostic Tools for Microsoft 14. The performance-monitoring tool built in to
Windows NT.” Microsoft Windows NT Server is called
Performance Monitor. For more information, see
9. The UNIX/Linux command-line utility that will
the section “Identifying Bottlenecks.”
display the current TCP/IP configuration is
ifconfig. For more information, see the sections 15. The utility in Microsoft Windows NT Server and
“Network Diagnostic Tools for UNIX” or Microsoft Windows 2000 Server that will display
“Network Diagnostic Tools for Linux.” the processes currently running on the network
server is called Task Manager. For more informa-
10. The TCP/IP utility used to test communications
tion, see the section, “Diagnostic Tools by
between the current network server and another
Operating System.”
server on the network is ping. For more informa-
tion, see the sections “Network Diagnostic Tools
for Windows NT” or “Network Diagnostic Tools
for Windows 2000.” Answers to Exam Questions
11. The name of the TCP/IP utility used to test name 1. B. A NIC with Wake-On-LAN technology pow-
resolution from a DNS server is nslookup. For ers up the computer when it receives a magic
more information, see the sections “Network packet. For more information, see the section
Diagnostic Tools for Windows NT” or “Network “Remote Troubleshooting for a Wake-On-LAN.”
Diagnostic Tools for Windows 2000.”
12 Serv+ Ch_07 7/10/01 9:57 AM Page 446
A P P LY Y O U R K N O W L E D G E
2. A. The processor utilization is greater than 70%, 8. C. A processor bottleneck can be eliminated by
indicating that the processor is the network server upgrading the existing processor to a faster proces-
bottleneck. For more information, see the section sor, adding another processor to the network server,
“Processor Bottlenecks.” and by moving applications on the network server
to a different network server. For more informa-
3. B. The command used to shut down a Novell
tion, see the section “Processor Bottlenecks.”
NetWare network server is down. For more informa-
tion, see the section “Novell NetWare Shutdown.” 9. B. The way to eliminate a memory bottleneck is
to install additional memory into the network
4. C. The indicators (statistics) that should be
server. For more information, see the section
monitored to determine whether the processor
“Memory Bottlenecks.”
is the bottleneck in the network server are per-
cent processor utilization and processor queue 10. D. The indicators (statistics) that should be
length. For more information, see the section monitored to determine whether the disk sub-
“Processor Bottlenecks.” system is the bottleneck in the network server are
percent disk utilization and disk queue length.
5. C. The Ping command is the TCP/IP utility used
For more information, see the section “Disk
to test network communications from one net-
Subsystems Bottlenecks.”
work server to another network server. For more
information, see the section “Troubleshooting 11. C. The Tracert command is the TCP/IP utility
TCP/IP Network Problems.” that reports the routers that a network packet
passes through going from one network server to
6. B. A POST Diagnostics Board is used when
another network server. The UNIX/Linux version
the network server will not boot. For more
of Tracert is traceroute. For more information,
information, see the section “Troubleshooting
see the section “Network Diagnostic Tools for
Boot Failures.”
Microsoft Windows NT” or “Network
7. B. The performance tests indicate a very high Diagnostic Tools for Microsoft Windows 2000.”
paging rate, which indicates high memory utiliza-
12. A. The very first thing to do as part of the net-
tion. The high percentage on disk subsystem uti-
work server shutdown process is to notify the
lization could indicate that the disk subsystem is
logged-on users that the network server is going
the network server bottleneck; however, the high
to be shut down shortly. For more information,
disk utilization is caused by the paging, which is
see the section “System Shutdown.”
caused by a shortage of memory in the network
server. For more information, see the section
“Memory Bottlenecks.”
12 Serv+ Ch_07 7/10/01 9:57 AM Page 447
A P P LY Y O U R K N O W L E D G E
13. A. The new SCSI disk drive has duplicated the 14. C. The ATA disk drives are not configured prop-
SCSI ID of another device on the SCSI channel, erly. One of the disk drives must be set to the
thus making both not appear on the list of recog- master role and the other must be set to the slave
nized SCSI devices. The SCSI bus being not cor- role because they are on the same ATA disk chan-
rectly terminated would cause all SCSI devices to nel. For more information, see the section
be unrecognized. The SCSI bus being too long “Troubleshooting an ATA Subsystem.”
would cause the system to be “flaky” in perfor-
15. B. The Destination host unreachable message
mance. Answer D is totally wrong, because SCSI
from a Ping command indicates that there is no
disk drives do not have master and slave settings.
way to get to the IP address of the Ping com-
For more information, see the section
mand. This indicates that the default gateway is
“Troubleshooting a SCSI Subsystem.”
not configured on the computer issuing the Ping
command. For more information, see the section
“Troubleshooting TCP/IP Network Problems.”
OBJECTIVES
OBJECTIVES OUTLINE
S T U DY S T R AT E G I E S
To prepare for the Disaster Recovery objectives of . If you have access to a server with a RAID con-
the Server+ exam, do the following: troller and hot-swap drives, set up the different
types of RAID arrays and experiment with drive
. Read the objectives at the beginning of
failures by pulling a drive from the array.
this chapter.
. Play the “what if?” game. Look around at the
. Study the information in this chapter.
computers that you have access to and figure
. Review the objectives again. out what you would do if something went
. Answer the Review and Exam Questions at the wrong. What if the hard drive fails? What if
end of the chapter and check your results. there is a flood? What if your office/house is
hit by lightning and your computer is dam-
. Use the ExamGear test engine on the aged? How would you continue to do busi-
CD-ROM that accompanies this book for ness? It doesn’t even need to be business
additional Exam Questions concerning systems. What if your PC hard drive crashes?
this material. How would you use the ExamGear CD included
. Review the Exam Tips available in various with this book to prepare for this exam? If you
places in this chapter. Make certain that you have no access to computers at this time, you
understand the information in the Exam Tip. If can do the same exercise with appliances.
you don’t fully understand the Exam Tip, How would you watch the Super Bowl if your
review the information in the chapter related TV is broken? What would you do with your
to this topic. food if your refrigerator stops working? The
object of this exercise is to put yourself in a
. If you are working for a company that has disaster-recovery frame of mind and get you
servers deployed, think about the different thinking about not only what is covered in this
types of things that could go wrong with them, chapter, but also the thousands of things that
and then think about what you would do in there is no way for us to cover here.
the event of a failure. Review your company’s
disaster-recovery procedure if available.
13 Serv+ Ch_08 7/10/01 9:59 AM Page 452
Recovery Plan! If you are ever Probably the best place to start a discussion of disaster recovery is
asked what is the most important with a definition of what is meant by disaster recovery. Disaster
EXAM
CREATING A DISASTER-RECOVERY
PLAN
. 7.1 Plan for disaster recovery
• Plan for redundancy (e.g., hard drives, power supplies,
fans, NICs, processors, UPS)
• Use the techniques of hot swap, warm swap and hot spare
to ensure availability
• Use the concepts of fault tolerance/fault recovery to create
a disaster recovery plan
• Develop disaster recovery plan
• Identify types of backup hardware
• Identify types of backup and restoration schemes
13 Serv+ Ch_08 7/10/01 9:59 AM Page 453
NOTE
But My Risk Analysis Book Says
need to perform a risk analysis.
There Are Three Parts! One thing is
very important to remember when you
do a risk analysis—and it’s not how
Risk Analysis many steps are in your process. It’s
“Why do I need to do a risk analysis?” is a common question asked be consistent. Make sure you use the
by technical staff. After all, isn’t that what the business people are same process for all your analysis to
for? If you’re a technical person, a lot of time the last thing you want ensure you get consistent results.
Everyone has his or her own ways to
to be doing is dealing with the processes, procedures, and paperwork
perform a risk analysis, and as long
involved in doing a risk analysis. And a risk analysis is not usually
as your method works for you, and
something a technical person can complete without some input gives you results you can work from, it
from the business people. As far as why a risk analysis is important, is a great tool. The four-step break-
think about driving your car. If you were to do a complete disaster- down is just suggested as a guideline.
recovery plan for your car, you would want to stock spare parts for
any component that might fail. So in addition to the spare tire, you
should also have a spare rear-view mirror, a spare steering wheel, and
a spare windshield. This is an extreme example to illustrate the
point. In your car, you know that a tire is the most likely thing to
fail that you might have a chance to repair. With your server, not
only do you need to know what components are most likely to
fail, are repairable, or can be dealt with by using high-availability
techniques, but you also need to understand what the value of the
information is.
So your first response is probably something along the lines of “Well
I already know what’s important, so why do I have to go through all
this analysis?” Besides, no one really understands this risk analysis
stuff anyway. Let’s take a look at what a risk analysis really is, and
discuss why you really need this.
13 Serv+ Ch_08 7/10/01 9:59 AM Page 454
A good risk analysis can be broken into the following four parts:
1. Identify your business processes and their associated infra-
structure. For example, communications is a business process
that involves a mail server, a network connection, and the end-
user workstation. Remote access is a process that might involve
modems, modem lines, a RAS server, and a network connec-
tion. It is absolutely critical that you involve someone from the
business in this part of the analysis. In a perfect world, the IT
department would be fully cognizant of all the processes and
procedures used by the business. The reality is frequently that
the IT department builds the infrastructure, but is not com-
pletely sure how the business is using it. And even if you do
know what the infrastructure is being used for, the value of the
business processes is generally not well understood. In many
cases the risk analysis will be the first time anyone thought to
place a value on processes and services.
2. Identify the threats associated with each of the business
processes and associated infrastructure. This could be any-
thing from a hard drive in the mail server to a failed modem
line on your RAS server to a network cable going bad. Try to
be specific and thorough. Be sure to list all the threats, and
not just the ones that you know you can address.
3. Define the level of risk associated with each threat. This
includes the likelihood of the threat occurring, as well as the
severity of the threat. You can do this by ranking the threats
on a scale of 1 to 5, where 1 is unlikely and 5 is very likely.
The severity scale would also be from 1 to 5, where 1 is minor
impact to the business and 5 is major impact to the business.
If possible, associate costs with each threat. This will make
justifying your disaster-recovery plan much easier.
4. Rank the risks based on severity and likelihood. One
method that works fairly well is to add the scores of the two
levels. The higher the score, the more important that risk is
to avoid. Low scores can generally be placed at the end of the
list and are usually the last risks to be planned for.
13 Serv+ Ch_08 7/10/01 9:59 AM Page 455
Now comes the fun part. You have now gotten a list of your poten-
tial threats. If you have done a thorough job, the odds are pretty
good that there are things on your list that you wouldn’t have
thought of if you just started with issues that you knew about. You
also now have a valuable tool for justifying the costs of your disaster-
recovery plan. It is much easier to justify that server cluster for the
company’s web server when you can show management that a
4-hour outage will cost the company $100,000 in lost business. If
you still have any question about why to invest your time and effort
in a risk analysis, think about this. When the risk that management
decided wasn’t worth the cost actually happens (and they inevitably
do) and they start looking for someone to blame it on, it will not be
the person who has the risk analysis documentation and manage-
ment’s refusal to make the necessary investment.
Now you need to determine how you can avoid these threats, and
provide the kind of availability your company needs. Let’s start by
talking about some of the more basic building blocks of disaster
recovery: the use of redundant components, clustering, scalability,
and high availability.
Understanding Redundancy,
Clustering, Scalability, and High
Availability
Because the topic of this book is servers, one of the best places to
start discussing some of the mechanisms for ensuring high avail-
ability is to look at some of the server components that can fail
from time to time. The most common component to be imple-
mented in a redundant fashion is that of the server hard drives.
Redundancy
Most of this discussion of redundancy focuses on hard drives, partic-
ularly RAID implementation. But following the sections on RAID,
the discussion focuses on some other components that are candi-
dates for redundancy.
13 Serv+ Ch_08 7/10/01 9:59 AM Page 456
RAID Types
An Easy Place to Ask Questions
TIP
This section starts by diving right into the most common redundant
This chapter spends a lot of time
technology you will encounter. RAID (redundant array of inexpen-
discussing RAID technologies
sive disks) technology enables you to configure your drive arrays to
EXAM
Data 1 Data 4 Data 7 Data 10 Data 2 Data 5 Data 8 Data 11 Data 3 Data 6 Data 9 Data 12
Disk Controller
FIGURE 8.1
RAID 0 uses data striping across all disks with
no parity. RAID 0 - Disk Striping
13 Serv+ Ch_08 7/10/01 9:59 AM Page 457
tion, because you can perform only one write to the array at a Disk 1 Disk 2
NOTE
What’s the Difference Between
2N drives to implement, where N is the number of drives Mirroring and Duplexing? There are
needed to make up the requisite storage capacity. To get 10GB two main differences between mirror-
of storage, for example, you need 2 × 10GB drives. ing and duplexing (see Figure 8.3).
First, with duplexing, because both dri-
á RAID 5, a striped array with parity. In a RAID 5 array, the
ves can be written to concurrently,
data is striped across the disks, much like RAID 0, but each there is a performance advantage
block of data has a parity bit written to a parity drive (see over a mirrored configuration. The sec-
Figure 8.4). There is not a specific drive dedicated to parity, ond advantage to duplexing is the
but for each block of data the parity information is stored on a increased redundancy. In a mirrored
different drive. RAID 5 requires a minimum of 3 drives to configuration, you can survive a drive
implement, but provides very fast read performance, and failure, but not a controller failure. In
medium write performance. Write performance in this config- a duplexed configuration, you can sur-
uration suffers due to the requirement to write parity informa- vive the loss of a drive or the loss of
tion for each block of data written. RAID 5 will also add a controller.
considerable overhead on the system when it needs to recover
from a failed drive, due to the need to re-create data using the
parity information before it can be read. RAID 5 is also the
most complex RAID configuration to design. RAID 5 needs
N + 1 drives to implement. To get 10GB of storage, for exam-
ple, you need 3 × 5GB drives.
13 Serv+ Ch_08 7/10/01 9:59 AM Page 458
FIGURE 8.3
With duplexing, each drive in the RAID 1 set is
connected to a different controller.
Data 1 Data 2 Data 3 Data 4 Data 1 Data 2 Data 3 Data 4
Disk 1 Disk 2
RAID 1 - Duplexing
Data 1 Data 3 Parity Data 7 Data 2 Parity Data 5 Data 8 Parity Data 4 Data 6 Parity
FIGURE 8.4
RAID 5 is a good compromise between
Disk Controller
the redundancy of mirroring and the cost
of striping. RAID 5 - Striping with Parity
Stripe Set 1
Data 1 Data 4 Data 7 Data 10 Data 2 Data 5 Data 8 Data 11 Data 3 Data 6 Data 9 Data 12
Disk Controller
Data 1 Data 4 Data 7 Data 10 Data 2 Data 5 Data 8 Data 11 Data 3 Data 6 Data 9 Data 12
FIGURE 8.5
In RAID 0/1 you get the performance of a
striped set combined with the redundancy of
Stripe Set 2 mirroring. This is an expensive solution in
RAID 0/1 - Mirrored Stripe Sets terms of the number of drives required.
13 Serv+ Ch_08 7/10/01 9:59 AM Page 460
R E V I E W B R E A K
Just to help get these RAID levels straight in your head, take a look
at Table 8.1, which compares the different RAID levels discussed in
this chapter.
TABLE 8.1
RAID L E V E L S
RAID Desc. Min.# of Max.# of Performance Fault
Level Drives Drives Tolerant
0 Striping 2 32 IO performance No
is very high
due to IO being
spread across
multiple drives.
No parity cal-
culation.
1 Mirroring 2 2 Twice the read Yes, 1
transaction drive
rate of single
disks, same
write trans-
action rate as
single disks.
1 Duplexing 2 2 Twice the read Yes, 1
transaction drive, 1
rate of single control-
disks, same ler
write trans-
action rate as
single disks.
5 Striping 3 32 Highest read Yes, 1
with data trans- drive
parity action rate.
Write perform-
ance slower than
striping due to
parity calcul-
ation.
IN THE FIELD
Now let’s talk about what you will generally see in the field. It is pos-
sible that you will see all these RAID configurations in the field, but
some are used more often than others. RAID 5 is probably the most
common configuration, because it gives you the best trade off
between fault tolerance and performance. It is also the most effi-
cient of the redundant configurations, requiring only 3 drives, and
then using the N + 1 drive cost, where N is the number of drives
needed for the array. The more drives in your array, the more efficient
a RAID 5 array is. With 3 drives, your overhead is 33%, but with 5
drives it drops to 20%. This is very attractive to cost-conscious com-
panies. You may also encounter mixed RAID implementations, where
the system volume is mirrored on a couple of small drives (RAID 1),
whereas the application/storage drives are in a RAID 5 configuration.
This is an excellent configuration for a mail server or a web server,
because it balances the performance and costs with the require-
ments of the operating system and the application. You will generally
not see RAID 0 or RAID 0/1, due to the lack of fault tolerance in the
former, and the expense of the latter.
Now that you are familiar with the different types of RAID, it’s time
to take a look at how to implement the RAID array—using hard-
ware or software.
RAID controller because the operating system can handle the RAID Disadvantages of Hardware Versus
stuff without having to spend the additional money. The other side Software Implementations One of
EXAM
thinks that the benefits of a hardware controller outweigh the mar- the areas that it is very easy to get
ginal cost savings of not buying the controller. confused about when talking about
the pros and cons of RAID imple-
mentations is the differences
between the two ways to support
your array. Be sure you know what
the pros and cons for both hard-
ware- and software-based arrays for
the exam.
13 Serv+ Ch_08 7/10/01 9:59 AM Page 462
Before getting too much further into discussing why to select hardware
NOTE
A Special Note for Windows NT or software RAID, you need to define exactly what is meant by both:
Users If you are using software RAID
on a Windows NT server, the system á Software RAID. Software RAID implementations are either
and boot partitions cannot be part of operating system–based, or they are RAID applications that
a stripe set. run on the server. You will most frequently see operating
system–based implementations in new server installs, but
may find RAID applications in use in older servers. Most
recent operating systems will support RAID 0 (striping),
RAID 1 (mirroring), and to a lesser extent RAID 5 (striping
with parity). In a software implementation, all array configura-
tion and management activities are controlled by a software
application and utilize the host processor for processing.
á Hardware RAID. Hardware RAID implementations are gen-
erally implemented using a host-based RAID adapter and ties
the array functions directly to the disk interface. This means
all array configuration and management activities are con-
trolled by the hardware controller and utilize the controller’s
CPU for processing. Different hardware RAID adapters offer
different combinations of RAID support, but most will sup-
port RAID 0, RAID 1, and RAID 5. Other RAID configura-
tions such as RAID 0/1 and RAID 3 are supported based on
the host controller’s capabilities.
NOTE
á Hardware RAID. Except for the fact that both are used for
NOTE
creating and maintaining arrays, hardware RAID implementa- Hardware Arrays Have Another
tions have little in common with their software counterpart. Advantage Hardware RAID imple-
mentations don’t need an operating
Unlike the software RAID, hardware arrays do not use server
system to boot. Because the array
resources such as memory or processor cycles. Because the
software is in the controller hardware,
server processor can execute applications while the array con-
you can access the software or work
troller’s processor is executing the array functions, the result is on the array even if you can’t get the
a much higher performance implementation. This is especially operating system up and running. In a
true during processor-intensive operations such as rebuilding software RAID, there can be a prob-
an array. Hardware controllers also have the additional benefit lem if the server can’t load the array
of being operating system independent. software when the disk that contains
the array software has failed. If you
Now that you have a good understanding of the intricacies of imple- are implementing a software RAID,
menting a RAID solution to ensure drive redundancy, take a look at you may find you need a separate
some of the other components in a server that are commonly used boot drive, which is not part of the
in redundant configurations. array. This adds another failure point
to the server drives.
Redundancy of Other System Components
A number of other components in your server can be configured in
a redundant manner, including the following:
á Power supplies. Most high-end servers will come with an
option to install additional power supplies. This allows the
server to shift the power load when a power supply fails.
á Cooling fans. Have you ever noticed how hot a server can get
after it has been running for a while? If you want to ensure your
server stays cool, you should look for a server that utilizes redun-
dant cooling fans. Not only does this allow the server to avoid
overheating in the event of a fan failure, but your server also ben-
efits from the additional cooling while all the fans are running.
Additional cooling fans are usually part of the server configura-
tion, and are not usually ordered as an option for the server.
13 Serv+ Ch_08 7/10/01 9:59 AM Page 464
The main point to take out of this section is that with the exception
of the system board, any critical components in your server can gen-
erally be configured in a redundant manner. And if your data is so
critical that you think you need an additional motherboard, it’s time
to think about clustering.
Introduction to Clustering
A cluster is a group of independent computers working together as a
single system. This system is used to ensure that mission-critical
applications and resources are as highly available as possible. (High
13 Serv+ Ch_08 7/10/01 9:59 AM Page 465
There are two different types of cluster models in the industry: the
shared-device model and the nothing-shared model.
á Shared-device model. In the shared-device model, applica-
tions running within a cluster can access any hardware
resource connected to any node/server in the cluster. As a
result, access to the data must be carefully coordinated. This
function is coordinated by a distributed lock manager (DLM),
which handles the arbitration of resources in the event of a
conflict. The coordination of the access adds significant over-
head to the cluster, as well as making it much more complex
to design, configure, and maintain. This model can also suffer
issues with scalability, because of the amount of coordination
required between nodes.
á Nothing-shared model. In the nothing-shared model, each
node has ownership of a resource, so there is no competition
for the resources. In this model, two servers share a disk
resource, but one of the servers owns the resource, and the
other server takes ownership of the resource only if the pri-
mary server fails.
the server with excess capacity. If you know you need 5GB of disk
space today, you can plan on needing more in the future. So be sure
to put the additional storage in the server at the outset. If you need
5GB today, put in 10GB. If your application will need a single
Pentium III 850MHz processor, put in a GHz processor, or dual
processors. In the long run spending the money at the outset is far
cheaper that trying to add components later. That said, the second
thing you need to do to ensure your server is scalable is to make sure
that you can add additional resources if necessary. If you are putting
a single Pentium III processor in your server, for example, it is a
good idea to purchase a server that can handle multiple processors,
so you can add an additional CPU if you need it. Don’t fill every
drive bay in the server if you can help it. Leave yourself the ability to
add a couple of 72GB drives if you need more storage in the future.
The question is how much extra capacity should you build in, and
how much additional capacity should you be able to add once the
server is installed? Although capacity planning is frequently more of
an art than a science until you can get some historical numbers to
plan from, it is usually a good idea to add an additional 25% to any
new server configurations to ensure scalability. You should also try to
leave some capacity for each of the components in the system. Don’t
fill all the card slots, memory slots, drive bays, or CPU slots if you
can help it. You never know when you might need to add additional
capacity, and taking out hardware to upgrade it is an order of mag-
nitude more difficult that just adding additional capacity.
The final concept covered relating to clustering and fault tolerance is Know Your Concepts Be sure you
TIP
TABLE 8.2
C O M M O N A VA I L A B I L I T Y N U M B E R S
Availability Requirement Allowable Downtime Per Year
99.999% 5 minutes
99.99% 53 minutes
99.95% 2 hours, 23 minutes
99.9% 8 hours, 45 minutes
99.5% 44 hours
99% 88 hours
Great, now you know about these components. How can you use
them as part of your disaster-recovery process? As part of your risk
analysis, you should have identified components in your server that
are at risk of failure. As part of the mitigation process for providing
steps to recover from this potential threat, you need to identify
replacement parts. You also need to make sure you understand the
underlying technologies involved. Do you need hot-swap hard drives
13 Serv+ Ch_08 7/10/01 9:59 AM Page 470
for that server, or can it be down for an hour while you replace the
drive from stock? This is where the cost-versus-benefit calculations
NOTE
NOTE
effective manner (this varies from server to server, and Where Should the Spare Parts Go?
depends on the cost of an outage) and start documenting As part of this process, you should be
the configuration. sure to identify a location for the stor-
age of spare parts. This should be a
5. Take any components that either cannot be implemented in a secure location ideally located in
fault-tolerant configuration or that for which a fault-tolerant close proximity to the server. You
configuration would be cost-prohibitive, and determine should also ensure that you are not
whether a spare part should be kept on hand in the event of the only person with access to these
an outage. If the answer is yes, prepare a list with part num- parts. You don’t want them stolen, but
bers and costs for those components. If the answer is no, try to you also don’t want to be flying back
from Aruba to open the parts cabinet
plan what you would do in the event that part fails. This could
when the server crashes. You also
be something like using a hot-spare server, making sure the
need to consider the environmental
server and its components remain under warranty, or making factors associated with storing your
sure you know where to get a new part. parts. These factors include keeping
6. Now that you have identified the hardware needed for your the parts stored at a safe tempera-
disaster-recovery plan, here comes the fun part: documenta- ture, away from static electricity or
moisture, and any of the other envi-
tion of the plan. Your disaster-recovery plan should include
ronmental factors covered in Chapter
documented contingencies for any of the threats you identified
6, “5.0—Environment.”
as part of the risk analysis. The purpose of this is threefold.
First, it gives management a warm, fuzzy feeling when they
realize you have your disaster-recovery plan in a folder next to
the server and are ready to go in the event of a problem. WA R N I N G
Second, if you are not available in the event of an outage, I’ve Got RAID 5, So I Don’t Need a
someone else can get the process started. Third, right now as Backup, Right? Wrong. You
should always back up your data to
you document all this, there are no alarms going off and no
a media that can be stored in a
VPs screaming at you to get the server fixed. This is an excel-
location that is physically separate
lent time to figure out what you’ll need to do in the event of
from the server itself. There are
an outage so that when the day comes that you are paged at 2 always those people who think they
a.m. you aren’t trying to remember that command to rebuild can save the cost and administra-
the RAID array. It will be part of your disaster-recovery plan. tion of a tape backup by putting in
redundant disks; but it is a mis-
7. After all this information has been documented, you need to
take, which they generally discover
place your orders and get ready to start configuring the server.
when the server crashes and they
You are ready to deal with a hardware failure.
can’t get to those highly redundant
disks.
Now that you have dealt with the potential hardware problems, take
a look at how to keep your data safe in the event of a disaster.
The following list identifies some of the tape technologies you will
encounter in the industry. Table 8.3 contains the storage and
throughput statistics for some of the more popular technologies, as
well as some that are a little less well known, for your reference.
á DDS (Digital Data Storage). DDS is the industry standard
for digital audio tape (DAT) formats. DAT tape backup drives
were very popular several years ago, and remain popular in
smaller servers. Their limited capacity compared to that of
newer technologies have prevented them from being used
heavily in an enterprise role. DDS drives tend to be inexpen-
sive, relative to the other, higher-capacity drives.
á DLT (Digital Linear Tape). DLT was the technology seen as
the successor to DDS drives toward the end of the 1990s.
DLT was a technology originally developed by Digital
Equipment Corporation but which is now available from a
variety of vendors. Its higher capacity and higher speed made
it popular for larger installations, and this is the technology
that the other manufacturers are competing against in high-
end installations.
á Mammoth. Developed in 1996 by the Exabyte Corporation,
Mammoth drives were to be the competitor to DLT from a per-
spective of capacity and performance. Although optimized for
superior reliability and exceptional price/performance, Mammoth
drives are not as popular as the more widely used DLT drive.
á Ultrium LTO (Linear Tape Open). Ultrium LTO technology
was developed jointly by HP, IBM, and Seagate to provide a
next-generation tape technology to compete in the high end of
the backup market. Although there hasn’t been wide adoption of
this high-end technology yet, Ultrium’s high capacity and
throughput performance make it a solid selection if you need the
horsepower. Expect to pay a premium for the new technology.
á AIT (Advanced Intelligent Tape). AIT was originally devel-
oped by Sony, and is available from a number of vendors. Like
the Ultrium technologies, AIT offers a high-capacity, high-
speed tape backup technology. AIT tape drives are known for
their unique MIC (Memory in Cassette) feature, which acts as
NOTE
When you get ready to pick your tape drive, you should determine
NOTE
how much data you will need to back up, and the amount of time you But My Data Won’t Fit on a Single
can allocate to the backup. You should avoid running backups during Tape If you have more data than you
can comfortably squeeze on a single
production hours, so be sure to determine what your maintenance
tape (and with the new AIT and
window will be for backups. Next, keeping in mind that the through-
Ultrium drives, that’s a lot of data),
put numbers listed in the table are theoretical, and your performance
you should investigate using a tape
will probably not be quite as high, select a technology that will enable library. A tape library is a backup sys-
you to complete your backup within that window. You should also take tem that contains multiple tapes (and
the backup method into account while making these calculations. sometimes drives) along with the
mechanics to allow it to automatically
change tapes. These units can scale
TABLE 8.3 up to multiple terabytes of data, so
you should be able to find one to
TA P E B A C K U P H A R D W A R E
accommodate your requirements.
Format Native Compressed Throughput Throughput
Capacity Capacity (Native) (Compressed)
(GB) (GB) (MBps) (MBps)
Travan-4 4 8 .5 1
AIT-2 25 50 2 4
DDS-4 50 100 6 12
DLT 8000 40 80 6 12
LTO 100 200 20 40
(Ultrium)
NOTE
It’s in the Software All these differ-
ent backup methods are pretty con-
Mammoth 2 60 120 12 24
fusing, especially if you are new to
some of the concepts. It took me for-
ever to keep the difference between
As noted in Chapter 5, “4.0—Proactive Maintenance,” there are
incremental and differential straight.
three different types of backups, as follows: Fortunately today’s tape backup soft-
á Full. A full backup will back up all the files marked for ware does most of the work for you. A
backup, and after the file has been backed up will set the good backup product will enable you
archive bit on the file to indicate it has been backed up. Full to select the backup method you want
to use, and will keep track of what
backups have the advantage of requiring a single tape or tape
tapes you will need when you do a
set to do a restore. If you want to perform a restore on
restore. Some will even enable you
Wednesday, you just use the Tuesday tape set. Because you are to select rotation strategies, and
backing up all the files, however, this is the slowest of the won’t allow you to overwrite a tape
backup methods. With a single-tape restore capability, how- unless it is ready to be taken out of
ever, this is the best method for recovering. Use full backups the rotation.
whenever possible for this reason.
13 Serv+ Ch_08 7/10/01 9:59 AM Page 474
IN THE FIELD
a weekly tape. These tapes are reused monthly. On the first of the
month, replace the tape that would ordinarily run with the monthly
tape, and perform a full backup. These tapes can be reused yearly.
And on December 31st you use the yearly tape, and put it aside
until the end of the following year. This theoretically gives you the
ability to go back as far as a year to recover your data.
Now, after you have all those tapes stored offsite, and you have a
year’s worth of tapes, how do you go about restoring a file that is
two months old? It’s simple. Go to the monthly tape that was used
just before the file you want was deleted or modified. If you want a
file that was deleted on April 13, and you run your monthly tapes
on the first of every month, you need the April tape to restore the
file. But wait, there’s a catch! What if you created and deleted the
file in January, and now it’s June? How do you get the data back?
The answer is you don’t. The G-F-S method is great for keeping
most of your data backed up, but it won’t get all your data. You will
be able to be sure of restoring only all your data during the week
the daily tapes are run.
Sounds horrible doesn’t it? But think about how many times you
have created a file, made a lot of changes, then saved it, and when
you went back in you accidentally deleted all your changes.
Because that file won’t be backed up until that evening, there is no
way to restore from tape. So the G-F-S method gives you one of the
better ways to keep your data archived, without the expense of 365
tapes a year.
IN THE FIELD
continued
NOTE
The final piece of the backup and restore puzzle is what you should
do with your tapes after you have data on them. As discussed in
Chapter 5, here are several schools of thought on this matter.
The easiest ways to store tapes is to place them in a fireproof box in
the server room. This protects the tapes from fire or water damage,
but keeps them close to the server for ease of tape management. If
you need to restore from tape, you just unlock the box, pull the
tape(s) you need, and perform your restore. You are at risk from a
catastrophic failure in the event of a tornado, flood, or major fire.
Another way to store tapes is to have them picked up by an offsite
storage facility. These facilities will come to your office once a day or
once a week and take your backup tapes to a secure warehouse.
Although his service incurs a cost, this is the most secure way to
store tapes, and it’s what most major corporations do.
Some other common methods for storing tapes include leaving them
on top of the server, locking them in the manager’s desk, having one
of the administrators take them home, and many other variations on
the same theme. The security, reliability, and tape-integrity issues
13 Serv+ Ch_08 7/10/01 9:59 AM Page 477
related to these types of tape storage methods make them very haz-
ardous to your continued employment. They just are not workable
in a business environment. Avoid them at all costs.
The following couple of exercises can help you get an idea of how to
test the pieces of your disaster-recovery plan.
STEP BY STEP
8.1 Testing Your Backups
In preparation for this test, you should create a directory on your
server with some test data in it. Make sure that it also has a directory
structure and custom permissions.
1. Retrieve a full backup tape from the preceding week’s
backup rotation. Insert it in your drive and after it has
completed loading, open your backup software application.
2. From the restore function of the tape backup software (this
varies from site to site, but all backup software has the same
basic features), select the test directory and select restore.
You should restore the data to a different location (different
drive if possible) than it was backed up from, to ensure that
your results are not contaminated by existing files.
3. When the restore has completed, check for the following:
• Was the data restored correctly? Is it all there; are
the files the correct size; and if you open a data
file, is the information correct?
• Was the directory structure restored correctly?
• Are the file and directory permissions correct?
4. If all the data was restored correctly, replace the tape in
the rotation and start planning for next week’s test. Be
sure to document your results. If your disaster-recovery
plan is ever audited, the auditor will want to see that your
tests have been successful.
You should test your tape backups at least once a week, and if
possible perform a full restore to a different machine quarterly, to
ensure the process is working, and that your administrators under-
stand the process.
13 Serv+ Ch_08 7/10/01 9:59 AM Page 479
STEP BY STEP
8.2 Testing Your RAID Configuration
Ideally this test should be done against a spare machine, or a lab
machine running a configuration similar to that of your production
equipment. You should never run a test of this type against a produc-
tion system. The purpose of this test is to verify that your recovery
procedure works.
1. After identifying a test machine, and verifying its RAID
configuration, remove a hard drive from the array. In a
system with hot-plug drives, this can be done by removing
the drive. If you do not have hot-pluggable drives, open
the case and remove the power cable from your drive.
2. Verify the system’s alarms on the drive failure. If the system
does not give you an alarm of some type, ensure that the
settings on the RAID controller are correct, and that any
software RAID-monitoring software is loaded on the
system. Then verify that your production systems are
configured correctly.
3. After the alarm has been verified, test the system. Log in,
run some applications, and do whatever you would nor-
mally do on the system. Ensure that the RAID did its job
and that all your data is available. Note any performance
degradation.
4. Bring the “failed” drive back into the system. Follow your
procedures for rebuilding the array. After the array has
been rebuilt, perform the same tests you performed in
step 3. Note any performance changes.
This test should be done at least twice a year, and any administrators
on the system should perform the test at least once to ensure that
they are familiar with the process and understand the documentation.
In some larger companies, yearly full tests of disaster-recovery plans
are done. This can include bringing hot sites (discussed later in the
chapter) online and ensuring business continuity is maintained,
13 Serv+ Ch_08 7/10/01 9:59 AM Page 480
Okay, now it’s show time. You have developed your plan, tested it,
and the big day is at hand. Your server just crashed. Let’s talk about
how you will implement your disaster-recovery plan. This is where
the rubber hits the road.
In the event of a failure, the first thing you need to do is determine
the cause of the outage. After you have identified the outage, it’s
time to figure out how to fix it. If you have a failed component, you
need to identify your hardware replacements. Each type of replace-
ment has different parameters you need to check to ensure it will
work in your system.
á Memory. For memory, you want to verify that the type of
memory is correct. (That is, is it error-correcting memory,
non-error-correcting, parity, and so forth?) You should also
check the speed of the memory (50 nanosecond, 60 nanosec-
ond, and so on), and finally check the form factor. Is it a
DIMM (dual in-line memory module) or the older SIMM
(single in-line memory module) type of memory? Also be
aware of any restrictions the server has on the way memory is
added. Some older servers required that memory be installed
in pairs, or in some cases sets of four.
á Disk drives. Disk drives are a little easier to replace than
memory, because there are fewer variables. You need to make
sure that the replacement drive is the same type (SCSI in most
cases, or possibly IDE), the same speed (most drives on the
market today are either 7,200 or 10,000 revolutions per
minute), and the same capacity. The size and speed are espe-
cially critical in a RAID array, because all the drives should be
the same size for the most efficient configuration.
13 Serv+ Ch_08 7/10/01 9:59 AM Page 481
á CPUs. You should always replace CPUs with a like model (for
example, Pentium II for a Pentium II). In a system with only
one CPU, you can often upgrade the speed of the CPU with-
out negatively affecting the system. If it is a multiprocessor
server, you need to keep the same speed and type as the rest
of the processors in the server.
á Network interface cards. NICs differ a little bit from the
previous items, because they interact with the operating
system through drivers. If you can manage it, you should
try to replace your NIC with the same make and model. If
you don’t mind updating drivers, however, you can replace a
NIC with a different make, model, or even vendor if neces-
sary. You just need to make sure that they match the media
type (10BASE-T Ethernet, 100BASE-T Ethernet, Token
Ring, and so forth).
Now that you understand all the facets of recovering from a server
outage, you need to consider one final topic: What do you do when
your facility is down? This could be due to a natural disaster such as
an earthquake or a flood, sabotage such as the bomb that went off in
the World Trade Center, or even just an extended power outage. You
need to find a place to resume your critical business activities. This
is where disaster-recovery sites come into the picture. Two types of
disaster-recovery sites are commonly used in the industry:
á Hot sites. A hot site is a commercial facility available for systems
backup. For a fee, these vendors provide a facility with server Hot and Cold For the exam, be
TIP
hardware, telecommunications hardware, and personnel to assist sure to know the difference
between a hot site and a cold site.
your company with restoring critical business applications and
EXAM
CHAPTER SUMMARY
You should now have a good handle on how to plan for and recover
KEY TERMS
from any disasters that might strike your server. As with just about
• Clustering anything in the IT industry, the key is good planning.
• DIMM This chapter began by looking at disaster-recovery planning. The
• Disaster-recovery plan importance of performing a risk analysis before starting your disaster-
recovery planning was discussed. This step is frequently overlooked as
• Distributed lock manager (DLM) people rush to start configuring servers, but for servers that contain
• Hardware RAID mission-critical data, this is an exercise that shouldn’t be skipped. In
summary, a good risk analysis contains the following steps:
• Hot spare
1. Identify your business processes and their associated
• Hot swap
infrastructure.
• Network interface card
2. Identify the threats associated with each of the business
• RAID processes and associated infrastructure.
• RAID 0 3. Define the level of risk associated with each threat.
• RAID 0/1 4. Rank the risks based on severity and likelihood.
• RAID 1
When you understand the risks, you can do a much better job of mak-
• RAID 5 ing sure you can recover from the most likely or most critical outages.
• Redundancy The discussion moved from risk analysis into some of the configura-
tions for high availability, starting with the different types of RAID
• Risk analysis
technologies available, and their pros and cons.
• Scalability
You also learned the pros and cons of software versus hardware
• SIMM RAID. Software RAID is less expensive, but adds a lot of overhead to
• Software RAID the system, because it uses the server processor and memory to per-
form its RAID functions. Although hardware RAID is more expen-
• Warm swap sive, it performs much better, is generally more flexible, and doesn’t
require the operating system to boot in order for it to be managed.
You also learned about other components that can be configured for
redundant operations. These include power supplies, cooling fans,
network interface adapters, CPUs, and UPSs.
13 Serv+ Ch_08 7/10/01 9:59 AM Page 483
CHAPTER SUMMARY
The discussion then turned to the advantages of using clustering
technologies to ensure system high availability. There are two types
of clustering, with different pros and cons. In the shared-device
model, applications running within a cluster can access any hard-
ware resource connected to any node/server in the cluster. This
model suffers from additional overhead and scalability issues, due to
the amount of coordination required between nodes. In the nothing-
shared model, each node has ownership of a resource, so there is no
competition for the resources. This model is more scalable and
higher performing than the shared device model, but can be more
expensive due to the lack of shared hardware.
Some of the advantages of clustering in general include the following:
á Fault tolerance
á High availability
á Scalability
á Easier manageability
CHAPTER SUMMARY
á Warm swap
á Hot spare / redundant components
You also learned about the different types of backups: full, differen-
tial, and incremental.
This rounded out the development of a disaster-recovery plan, and
from there the discussion focused on how to test a disaster-recovery
plan. This discussion dealt with some guidelines for testing, but rec-
ognized that detailed testing will vary from plan to plan. The key
concepts for testing a disaster-recovery plan include the following:
á Checking the recovery-plan documentation
á Doing a “dry run” of each of the components of your plan
á Testing your notification processes
á Checking the locations of any hot-spare equipment or servers
á Verifying your support contracts
13 Serv+ Ch_08 7/10/01 9:59 AM Page 485
CHAPTER SUMMARY
One other thing that you can add if applicable is to verify that your
hot or cold site is ready to accept the transfer of operations.
And the final section of the chapter dealt with what to do when dis-
aster strikes. In the event of a component failure in your server, you
need to make sure that you have a like component on hand, or that
you can get one quickly. Some components will allow you more flex-
ibility in their replacement than others; and this section covered the
parameters of the common replacement parts.
The final topic covered was the use of alternative facilities in the
event of a major outage. You can use two types of facilities in this
event. A hot site is a commercial facility available for system back-
ups. A cold site (also known as a shell site) is a facility that has been
prepared to receive computer equipment in the event that the pri-
mary facility is destroyed or becomes unusable.
You have made it to the end of the discussions on disaster recovery.
This is an area of the business that is often neglected, and by having
the ability to design and document a thorough, structured, and doc-
umented disaster-recovery plan, you can ensure that your servers will
remain highly available. If you don’t, you may find that your server
isn’t the only thing that isn’t highly available; you may find your job
isn’t either.
13 Serv+ Ch_08 7/10/01 9:59 AM Page 486
A P P LY Y O U R K N O W L E D G E
Exercises 4. Rank the risks based on severity and likelihood.
One method that works fairly well is to add the
8.1 Performing a Risk Analysis and scores of the two levels. The higher the score, the
Creating a Disaster-Recovery Plan more important that risk is to avoid. Low scores
can generally be placed at the end of the list, and
The purpose of this exercise is to give you practice per- are usually the last risks to be planned for.
forming a risk analysis and creating a disaster-recovery
plan for a relatively simple server: your home PC. 5. Starting with the most critical risk, and working
your way down through at least the first five risks
Estimated Time: 60 minutes. (do more if you want the additional practice)
Required Equipment: Personal computer. devise a way to either counter or recover from the
threat. If your number one threat is a virus, pur-
1. Identify your business processes and their associ- chasing virus protection and updating the virus
ated infrastructure. In this case, think about what signatures weekly could be your counter. Think
you use your PC for. Your “business process” may about how you could use some of the technolo-
be email, writing letters, online chats, preparing gies in the chapter (redundancy, spare parts,
for the Server+ exam, or whatever you use your RAID, and so on) to reduce the threats.
home PC for.
6. Associate costs with each disaster-recovery item.
2. Identify the threats associated with each of the How many are worth the investment? Include in
business processes and associated infrastructure. your calculations things such as the cost of a
What would happen if your modem were to fail? resume that you submit to a potential employer
Your hard drive? What if you caught a virus? Try that has a virus in it, and costs you the job, or the
to think of all the possible threats to your PC. cost of failing the Server+ exam if your monitor
Include things that you would consider low risk, fails, in your decision as to whether it is a worth-
as well as high-risk threats. while investment.
3. Define the level of risk associated with each 7. Document your findings.
threat. This includes the likelihood of the threat
occurring, as well as the severity of the threat.
You can do this by ranking the threats on a scale 8.2 Working with Tape Backup Methods
of 1 to 5, where 1 is unlikely and 5 is very likely. The purpose of this exercise is to give you practice
The severity scale would also be from 1 to 5, working with the different tape backup methods for
where 1 is minor impact to the business and 5 is backups and restores.
major impact to the business. In this exercise
(unless you use your home PC for a business), it Estimated Time: 60 minutes.
will be difficult to associate a direct cost for the Required Equipment: 3 × 5 index cards.
threat. In a true risk analysis, you should associ-
ate costs with each threat. 1. Take 7 index cards and label them with the days
of the week. These are your tapes.
13 Serv+ Ch_08 7/10/01 9:59 AM Page 487
A P P LY Y O U R K N O W L E D G E
2. For the purpose of this exercise, assume your 8.3 Creating RAID Configurations
backup rotation begins on Monday. Label each
index card Full Backup. The backups run at 10 The purpose of this exercise is to give you practice cre-
p.m., and take 2 hours to complete. ating RAID arrays. If you have the correct equipment
(a computer with four hard drives), you can perform
3. Assume you have successful backups throughout this as a live exercise. If you do not, use the index
the week, and it’s Saturday morning, around 11 cards.
a.m. Your server just crashed, and you need to do
a full restore to a spare server. How many tapes Estimated Time: 60 minutes.
will you need to complete the restore? Which Required Equipment: 3 × 5 index cards.
tape(s) will you need?
1. Take 4 index cards and label each 10GB. It repre-
4. Now leave the Monday card labeled Full, and sents a disk drive for your RAID array.
label the rest of the cards with Incremental. The
backups still run at 10 p.m., and the Monday 2. You need to create a RAID 0 array with your
backup takes 2 hours, but the remaining backups drives. What is the minimum number of drives
take approximately 1 hour. you will need? How much effective disk space
will that provide? If you use all 4 drives for this
5. Assume you have successful backups throughout array, how much effective disk space will be
the week, and it’s Thursday evening, around 8 available?
p.m. Your server just crashed, and you need to do
a full restore to a spare server. How many tapes 3. Using the same 4 drives, create a RAID 1 array.
will you need to complete the restore? Which What is the minimum number of drives needed?
tape(s) will you need? How much effective disk space will that array
provide? If you use all 4 drives for this array, how
6. Once again, leave the Monday card labeled Full, much effective disk space will be available? (Hint:
and label the rest of the cards Differential. The You can create 2 RAID 1 arrays for this step.)
backups still run at 10 p.m., and the Monday
backup takes 2 hours, but the remaining backups 4. Using the same 4 drives, create a RAID 5 array.
take approximately 45 minutes. What is the minimum number of drives needed?
How much effective disk space will that array
7. Assume you have successful backups throughout provide? If you use all 4 drives for this array, how
the week, and it’s Wednesday evening, around much effective disk space will be available?
10:15 p.m. Your server just crashed, and you
need to do a full restore to a spare server. How 5. Using the same 4 drives, create a RAID 0/1 array.
many tapes will you need to complete the restore? What is the minimum number of drives needed?
Which tape(s) will you need? How much effective disk space will that array
provide? If you use all 4 drives for this array, how
much effective disk space will be available? (Hint:
You can create 2 RAID 1 arrays for this step.)
13 Serv+ Ch_08 7/10/01 9:59 AM Page 488
A P P LY Y O U R K N O W L E D G E
6. If you think you need additional practice with 5. You’re the LAN administrator for Jeff ’s
RAID arrays, try this exercise using different Automotive Repair Ltd. You have been asked to
numbers of drives. configure the backup software for backing up the
customer database server. If the goal of the back-
ups is to allow for the fastest possible restore,
what type of backup should you use? How would
Review Questions you do it if the goal was to allow for the fastest
1. You are the server administrator for Exponent backup time? What type of backup provides a
Mathematicians and you have just taken over good compromise between the two?
responsibility for a server running a RAID 1
drive configuration. What are the benefits and
drawbacks to this RAID method? What are the
benefits and drawbacks to using a different RAID
Exam Questions
configuration? 1. You are the network administrator for Wild
Widgets Inc. and you need to build a front-end
2. You have been tasked to evaluate your company’s
web server to provide transactional data for
existing disaster-recovery plan, and see whether
e-commerce customers. This server will be part
adding a hot or cold site to the plan would make
of a redundant cluster of servers, and you need to
sense. What are the uses of each, and why would
design it so that it can deliver data from the disk
you select one over the other?
as quickly as possible. Because it is part of a
3. You are the administrator for John’s House of redundant cluster, the data integrity/availability
Tires, and you have been asked to evaluate the of the server is not a main concern. Performance
possibility of using clustering technologies for is critical.
your mission-critical web site. What are the two
Which RAID type should you select?
types of clustering, and how do they differ?
A. RAID 0
4. You are building a new file server for your corpo-
rate headquarters and want to make sure that you B. RAID 1
can recover from a disk failure as quickly as possi-
C. RAID 5
ble. What kind of disks should you use? What are
your other options? D. RAID 0/1
13 Serv+ Ch_08 7/10/01 9:59 AM Page 489
A P P LY Y O U R K N O W L E D G E
2. You are the network administrator for Wild Which of the following are benefits to running a
Widgets Inc. and you need to build a database clustered environment?
server for the accounting department. The man-
i. Fault tolerant
ager of the accounting department is concerned
about the integrity of his data, but he is also very ii. High availability
cost conscious. He wants to implement this as iii. Scalability
efficiently as possible.
iv. Simple configuration
Which RAID type should you select?
v. Easier manageability
A. RAID 0
A. i only
B. RAID 1
B. i, ii, iii, and iv
C. RAID 5
C. i, ii, iii, and v
D. RAID 0/1
D. v only
3. You are the network administrator for
Wild Widgets Inc. and you need to build 5. You are the lead engineer for Little Faith
an e-commerce server for your company’s Enterprises, and you have guaranteed 99.99%
online store. This application is absolutely availability for the server cluster that you just
critical to the company’s success, so the implemented.
need for redundancy is absolute. Because Because the application must be available 24 × 7
thousands of customers will be placing × 365, how much downtime enables you to still
orders and viewing the catalog, however, meet your availability numbers?
performance is equally important.
A. 5 minutes
Which RAID type should you select?
B. About an hour
A. RAID 0
C. More than 2 hours
B. RAID 1
D. 30 minutes
C. RAID 5
6. You are the network administrator for the Hittem
D. RAID 0/1 Boxing Glove Corporation. You are installing a
4. You have been asked to deploy a clustered server new server and want to configure it as redun-
as part of your job as Widgets, Inc.’s server dantly as possible.
administrator. Your manager has asked you to
assist him in justifying the expense.
13 Serv+ Ch_08 7/10/01 9:59 AM Page 490
A P P LY Y O U R K N O W L E D G E
Which of the following are not redundant com- i. Flexible
ponents in most servers?
ii. Inexpensive
A. CPU
iii. Less impact to system performance
B. Motherboard
iv. More redundant than software RAID
C. Power supply
A. i and ii
D. Network interface card
B. ii and iii
E. Hard drives
C. i, iii, and iv
7. You manage the servers for the ACME
D. iv only
Screwdriver Corporation. You are in the process
of developing a disaster-recovery plan and you 9. You are the LAN administrator for UR Write
have just started your risk analysis. Publishing, a technical publishing company. You
are in the process of implementing a RAID con-
Which of the following are part of a good risk
figuration using software RAID and your manager
analysis?
wants to know why you selected software RAID.
i. Justifying costs
Which of the following are advantages of hard-
ii. Identifying threats ware RAID?
iii. Ranking threats based on cost A. Flexible
iv. Identify business processes B. Inexpensive
v. Deciding which RAID level to use C. Less impact to system performance
A. i and v D. More redundant than hardware RAID
B. ii, iii, and iv 10. You are the data center manager for Mount Abora
Consulting, an independent consulting company,
C. ii and v
and you have been asked to specify the backup
D. i, ii, iii, iv, and v media for the new data warehouse application.
8. You are the LAN administrator for UR Write This application will eventually have between 100
Publishing, a technical publishing company. You and 150GB of data, and needs to be backed up
are in the process of implementing a RAID config- after hours, to reduce the impact on users.
uration using hardware RAID and your manager
wants to know why you selected hardware RAID.
Which of the following are advantages of hard-
ware RAID?
13 Serv+ Ch_08 7/10/01 9:59 AM Page 491
A P P LY Y O U R K N O W L E D G E
What tape technology should you recommend? i. Highly redundant
A. DDS-4 ii. Very efficient use of hardware
B. LTO iii. Highest performance RAID configuration
C. DLT iv. No degradation in performance when a
drive fails
D. Mammoth
A. iii and iv
E. Floppy disks
B. ii and iii
11. You are the owner/operator for Phil’s Vacuum
Cleaner Repair Service. In support of your busi- C. i and iv
ness, you have five PCs and a small server that
D. i only
stores your accounting information.
13. You have just set up a tape backup rotation and
Which of the following tape technologies would
are using the G-F-S method.
not be appropriate in your situation?
How many tapes will you need if you want to
i. DDS-4
have a full archive of tapes for two years?
ii. LTO
A. 20
iii. DLT
B. 21
iv. Mammoth
C. 40
v. Floppy disks
D. 42
A. ii, iv, and v
14. You are the lead administrator for the Got Bux
B. i only Savings and Loan, and you have just completed
your disaster-recovery plan.
C. i and v
Which of the following are part of a good disaster-
D. ii and v
recovery test?
12. You are the lead engineer for Zanzibar Dream Ice
i. Have someone check your documentation
Cream and you have just implemented mirroring
for completeness.
on your primary production server. Your manager
wants to know why you selected mirroring for ii. Cause one of your servers to fail and then
fault tolerance. recover from it.
Which of the following statements about mirror- iii. Inventory your spare parts.
ing would not be accurate in your justification?
iv. Identify a cold site to relocate your
servers to.
13 Serv+ Ch_08 7/10/01 9:59 AM Page 492
A P P LY Y O U R K N O W L E D G E
A. i 2. A hot site is a facility provided by a outside vendor
and contains all the facilities you need for business
B. i and ii
resumption. It is the more expensive of the two
C. i and iii possibilities. The cold site is just a facility that can
D. ii and iv hold your equipment in the event of an outage.
Although this is a less expensive solution, it will
15. You are the lead administrator for the Got Bux take longer to resume operations. For more infor-
Savings and Loan, and you are configuring your mation, see the section “Restoring Your Server.”
server with as much redundancy as possible. You
would like the server to continue to run without 3. The two clustering models are the shared-device
intervention in the event of a component failure. and the nothing-shared models. In the shared-
device model, applications running within a clus-
Which of the following components could fit this ter can access any hardware resource connected to
description? any node/server in the cluster. This model can
i. Power supply also suffer issues with scalability and perfor-
mance, but is more efficient from a hardware per-
ii. Fan spective. In the nothing-shared model, each node
iii. Motherboard has ownership of a resource, so there is no com-
petition for the resources. This model requires
iv. Video adapter more hardware to implement, but is more effi-
A. i cient and scalable, because there is not resource
contention. For more information, see the section
B. i and ii “Introduction to Clustering.”
C. i and iii 4. If speed is of the essence, you should use hot-
D. ii and iv swap drives, which you can add and remove from
a computer while the computer is running. If you
could handle a greater downtime, you could use
either warm-swap or hot-spare drives. Warm-
Answers to Review Questions swap drives requires that processing be halted
1. RAID 1 provides excellent redundancy, but is until the drive is replaced. A hot-spare drive
very inefficient. RAID 0 is very high perfor- would require the system be shut down while it is
mance, but offers no redundancy. RAID 5 offers replaced. For more information, see the section
the best combination of performance, efficiency, “Understanding Hot Swapping, Warm Swapping,
and redundancy, but requires more drives than and Hot Spares.”
the existing RAID 0 implementation. RAID 0/1
is the most inefficient, but if configured correctly
offers high performance and complete redun-
dancy. For more information, see the section
“RAID Types.”
13 Serv+ Ch_08 7/10/01 9:59 AM Page 493
A P P LY Y O U R K N O W L E D G E
5. For the fastest restore time, you would want to do 5. B. A 99.99% availability guarantee allows you
full backups every night. That would enable you less than an hour of downtime per year. 53 min-
to always restore from a single tape. For the short- utes to be more exact. For more information, see
est backup times, you would want to do a full the section “Understanding Redundancy,
backup on a non-business day and then incremen- Clustering, Scalability, and High Availability.”
tals for the rest of the week. To restore you would
6. B. In most servers you can get redundant CPUs,
need the full backup plus any subsequent incre-
NICs, power supplies, and hard drives.
mental tapes. And a good compromise would be a
Motherboards cannot be implemented redun-
full backup one night, with a differential the rest
dantly. For more information, see the section
of the week. You can restore from just two tapes,
“Redundancy of Other System Components.”
and you don’t have the additional overhead of a
full backup every night. For more information, 7. B. Deciding which RAID level to use is part of
see the section “Backing Up Your Data.” your disaster-recovery planning, which is probably
when you’ll be justifying the costs as well. For
more information, see the section “Risk Analysis.”
A P P LY Y O U R K N O W L E D G E
11. A. The Ultrium LTO and Mammoth technolo- 14. C. You should always have your documentation
gies are overkill for the environment described, double-checked, and be sure you have the appro-
and would be prohibitively expensive for a small priate spare parts on hand. Identifying a cold site
operation of this type. Floppy disks are not an should be part of the planning process, and you
appropriate backup technology anywhere. For should never induce a problem in a production
more information, see the section “Backing Up system just to see whether your process works.
Your Data.” What if it doesn’t work, and your server is down
for a week? For more information, see the section
12. B. RAID 1 (mirroring) is highly redundant, and
“Testing the Plan.”
there is no degradation in performance during an
outage due to the lack of parity calculations 15. B. Fans and power supplies can usually be con-
found in RAID 5. Mirroring is not an efficient figured redundantly. Motherboards and video
way to use hardware, due to the 100% overhead adapters usually cannot, although an additional
in disks, and it does not perform as well as RAID video card could be installed and the cutover
0. For more information, see the section “RAID done manually. For more information, see the
Types.” section “Redundancy of Other System
Components.”
13. D. To archive 2 years of data, you will need 24
monthly tapes, 8 weekly tapes, 8 daily tapes, and
2 yearly tapes, totaling 42. For more information,
see the section “Backing Up Your Data.”
1. www.labmice.net/disaster.htm 5. www.gedisasterrecovery.com/index.html
2. www.microsoft.com/windows2000/library/ 6. http://developer.novell.com/research/
operations/fileandprint/recovery.asp appnotes/1999/septembe/03/index.htm
II
P A R T
FINAL REVIEW
Fast Facts
Practice Exam
14 Serv+ Ch_P2 7/10/01 10:01 AM Page 496
15 Serv+ Ch_FF 7/11/01 10:06 AM Page 497
SERVER+ CERTIFICATION
15 Serv+ Ch_FF 7/10/01 10:02 AM Page 498
The following list runs down the job dimensions á An enterprise server is a network server that can
discussed in this book and that CompTIA uses to support a very large number of users or a very
arrange the objectives for the Server+ exam: large application, such as a major database.
á 0.0—Planning and Specifications á A server appliance is a network server that is
á 1.0—Installation designed to provide one service to the network
and take a relatively small amount of space.
á 2.0—Configuration
á A gateway server provides a communications path
á 3.0—Upgrading between two systems that use a totally different
á 4.0—Proactive Maintenance communications system, such as an Ethernet
LAN communicating with an IBM mainframe
á 5.0—Environment that is using SNA.
á 6.0—Troubleshooting and Problem á A router connects several network segments
Determination together and sends data packets from one seg-
á 7.0—Disaster Recovery ment to another based on the network address of
the data packet.
á A bridge segregates data frames between two sec-
tions of the same network and passes packets from
0.0—PLANNING AND one section to another based on the hardware
SPECIFICATIONS addresses of the source and destination computers.
á A firewall connects a corporate LAN to the
This job dimension is not one of the job dimensions Internet while restricting access from the Internet
listed on the Server+ Exam Blueprint from CompTIA. to the corporate LAN based on source IP address
However, there are items on CompTIA’s “Server+ and/or protocol type.
Concepts and Skills” document that aren’t included in
any of the job dimensions of CompTIA’s Server+ Exam á A Network Address Translation (NAT) server con-
Blueprint. These items are covered in the author- nects a corporate LAN to the Internet. The corpo-
created Job Dimension 0, Planning and Specifications. rate network generally uses private IP addressing
The main points are as follows: and the NAT server has a limited number of valid
IP addresses that it uses to translate the private
á There are two types of computer networks: peer addresses to addresses that are valid on the Internet.
to peer, and server centric (aka client/server).
á A proxy server connects a corporate LAN to the
á Peer-to-peer networks are useful for only relative- Internet. A proxy server is often the only computer
ly small networks of 20 computers or fewer. in the corporation that is actually attached directly
á A workgroup server is a network server that can to the Internet. All requests for Internet access are
support about 20 users. actually directed to the proxy server. The proxy
server retrieves the requested information, provides
á A departmental server is a network server that can it to the client computer who requested it, and
support up to 300 users. may also cache it on its own disk drives for future
access by all the clients on the corporate LAN.
15 Serv+ Ch_FF 7/10/01 10:02 AM Page 499
á A database server runs an application known as a á A Domain Name Service (DNS) server provides
database management system. The database server TCP/IP name resolution for computers on the
stores databases on its hard disk drives. Client TCP/IP network.
computers send requests or queries to the database
á A Windows Internet Naming Service (WINS)
server, the database server processes the request
server provides NetBIOS name resolution for
and retrieves the data from the database on its
Microsoft computers on a TCP/IP network.
hard disk drives, and then the database server
sends the results to the query back to the client. á A Dynamic Host Configuration Protocol
(DHCP) server provides automated TCP/IP con-
á A client/server is actually a classification for a
figuration for client computers on the network.
type of relationship between network servers and
their clients. The classic example of a client/server á A web server stores and provides web pages to
relationship is the database server. client computer web browsers.
á An application server is a server that provides a á A video server stores digitized video and provides
major service to the clients on the network. Good it on demand to client computers on the network.
examples of application servers include email,
á A CD-ROM library server provides access to the
database, and web servers.
information stored on many CD-ROMs to
á A mail server provides electronic mail services to clients on the network.
client computers on the network.
á A directory services server provides a searchable
á An FTP server provides a repository of files that database of objects (users, groups, computers,
can be accessed by and copied to client computers printers, and so forth) on the network that can be
on the network using the File Transfer Protocol. used by client computers on the network.
á An SNA server provides communications and á The Intel Pentium Pro processor supports up to 4
character translation between a LAN and an processors in a network server. It is available in
IBM mainframe. speeds from 150MHz to 200MHz. Its Level 2
cache runs at the same speed as the processor.
á A remote access server enables client computers
to access the network via modems attached to á The Intel Pentium II processor supports up to 2
telephone lines. processors in a network server. It is available in
speeds from 233MHz to 450MHz. Its Level 2
á A file and print server provides access to files and
cache runs at one-half the speed of the processor.
network printers to clients on the network. This
is the classic network server role. á The Intel Pentium II Xeon processor supports up
to 4 processors in a network server. It is available
á A terminal server provides access to desktop
in speeds of 400MHz and 450MHz. Its Level 2
applications to client computers on the network
cache runs at the same speed as the processor.
that don’t have the resources to run these desktop
applications themselves. (These clients are often á The Intel Pentium III processor supports up to 2
called thin clients.) processors in the network server. It is available in
speeds from 450MHz to 1130MHz. Its Level 2
á A fax server provides client computers on the
cache runs at the same speed as the processor.
network access to both incoming and outgoing
facsimiles.
15 Serv+ Ch_FF 7/10/01 10:02 AM Page 500
á The Intel Pentium III Xeon processor supports up á Symmetric multiprocessing (SMP) allows multi-
to 4 processors in the network server. It is available ple processors to be used cooperatively in a net-
in speeds from 500MHz to 1000MHz. Its Level 2 work server.
cache runs at the same speed as the processor.
á The memory bus connects main memory to the
á The Intel Pentium 4 processor supports only one processor of the network server. The memory bus
processor in the network server. It is available in operates at 66MHz, 100MHz, or 133MHz. The
speeds from 1300MHz to 1500MHz. Its Level 2 memory bus is also called the front-side bus.
cache runs at the same speed as the processor.
á The system bus allows expansion of the input and
á The future Intel Itanium processor is Intel’s first output capabilities of the network server by inser-
64-bit processor and it will support up to 4 proces- tion of expansion boards (video, disk controller,
sors in a network server. It will initially be available network adapter, and so on) into slots on the bus.
in speeds of 733MHz and 800MHz. It will have
á Table 1 summarizes system buses that have been
both Level 2 and Level 3 caches available.
used in computers over the past few years.
á Level 1 processor cache is built on the chip with
the processor. The relatively small Level 1 cache has
two parts: the data cache and the instruction cache. SUMMARY TABLE 1
Each of these is usually 8KB or 16KB in size. SYSTEM BUSES
á The memory used in both Level 1 and Level 2 Bus, Bus Width Bus Speed Bandwidth
cache is static ram (SRAM). SRAM is much Year Introduced (Bits) (MHz) (MBps)
faster than dynamic RAM (DRAM), which is ISA (8-bit), 1981 8 4.77 4.77
used as main system memory; however, it also ISA (16-bit), 1984 16 8 16
costs much more.
EISA, 1988 32 8 32
á Level 2 processor cache acts as a buffer between
Micro Channel, 1987 16 8 16
main memory and the processor. Level 2 cache,
like Level 1 cache, consists of SRAM. On Micro Channel, 1987 32 8 32
Pentium processors and higher, Level 2 cache is VESA Local Bus, 1992 32 33 132
on the chip with the processor. Level 2 cache PCI, 1992 32 33 132
ranges in size from 128KB to 2MB.
PCI, 1992 64 33 264
á When writing to cache memory, there are two PCI 2.1, 1995 64 66 528
different methods that are used: write thru and
PCI-X 1.0, 1999 64 132 1056
write back.
AGP, 1996 32 66 264
á In the write thru cache method, data is written
back to cache and main memory at the same time. AGP (x2), 1996 32 132 528
á The Extended Industry Standard Architecture á Extended data out (EDO) RAM has access speeds
(EISA) bus, which is compatible with ISA measured in nanoseconds (typically fewer than
adapter cards, was designed by nine IBM 100 nanoseconds).
competitors in response to IBM’s Micro
á Synchronous dynamic RAM (SDRAM) runs at
Channel Architecture bus.
bus speeds, which are measured in megahertz.
á The Peripheral Component Interconnect Common speeds are 66MHz, 100MHz, and
(PCI) bus is a local bus that was designed by 133MHz.
Intel Corporation.
á Error Correction Code (ECC) is a sophisticated
á The PCI bus can support a technology known as method of memory data protection that can
PCI hot swap, which allows a PCI adapter to be detect and correct single-bit errors. RAM that has
installed, removed, or replaced without shutting ECC capabilities is known as ECC memory.
down the network server.
á Memory that does not have ECC capabilities is
á The PCI-X (PCI Extended) bus is an enhance- known as non-ECC memory.
ment to the PCI bus and has a higher bus
á Buffered EDO memory has better performance
transfer speed.
than regular EDO memory because of the ampli-
á The AGP (Advanced Graphics Port) bus was fied control signals provided by buffering.
designed by Intel Corporation as a high-speed
á Registered SDRAM memory has better perfor-
local bus to be used primarily for video graphics.
mance than regular SDRAM because of the
á Intelligent Input/Output (I2O) technology is amplified control signals provided by registering.
designed to remove the burden of performing
á Integrated Drive Electronics (IDE) is a disk sub-
input and output operations from the network
system that supports up to two channels with up
server processor. It is also designed to implement
to two disk drives on each channel, for a total of
device driver independence from the operating
four disk drives.
system.
á IDE is also known as ATA (Advanced Technology
á Memory interleaving is a technique used to
Attachment) by the American National Standards
speed up memory access by dividing memory in
Institute.
sections, which can be accessed simultaneously.
á IDE/ATA disk drives can transfer data at a rate of
á A single in-line memory module (SIMM) can
3.3 to 8.3 megabytes per second.
transfer data 32 bits at a time, and they are
usually installed in pairs (because most data á Enhanced IDE (EIDE), also known as ATA-2,
buses are 64 bits wide). can transfer data at a rate of 11.3 to 16.5
megabytes per second.
á A dual in-line memory module (DIMM) can
transfer data 64 bits at a time, and hence they á ATA 33, also known as Ultra ATA 33, can trans-
can be installed singly. fer data at up to 33 megabytes per second.
15 Serv+ Ch_FF 7/10/01 10:02 AM Page 502
á ATA 66, also known as Ultra DMA 66, can á The standard SCSI channel width is 8 bits (1 byte
transfer data at up to 66 megabytes per second. or 1 character).
á Ultra DMA 100 can transfer data at up to 100 á Each end of a SCSI channel must be terminated.
megabytes per second.
á Each device on a SCSI channel must have a
á The Small Computer System Interface (SCSI) unique SCSI ID.
is a bus controller that can have many devices
á There is a Wide SCSI channel standard, which is
attached to a single SCSI channel.
16 bits wide (2 bytes or 2 characters).
á Table 2 summarizes the many versions of the
á Characteristics of the SCSI-1 standard include a
SCSI standard.
bus speed of 5MHz, a bus width of 8 bits, and a
data-transfer speed of 5MBps.
á 100BASE-F Ethernet uses fiber-optic cable. It has á Network interface cards (NICs) are installed in
a data-transfer rate of 100Mbps, uses baseband network servers to give them the capability to
transmission, and has a maximum cable length of attach to a network. The speed and width of the
2 kilometers. system bus determine the performance (data-
transfer speed) of the NIC.
á Gigabit Ethernet has a data rate of 1000 megabits
(1 gigabit) per second. It can use UTP cable for a á Adapter teaming is a technique of using multiple
distance of 100 meters. It can also use fiber-optic NICs in a network server to provide adapter
cable for a distance up to 2 kilometers. fault tolerance, adapter load balancing, or port
aggregation.
á 1000BASE-T (IEEE 802.3ab) uses Category 5 or
higher UTP cable. It has a data-transfer speed of á Adapter fault tolerance is achieved when two
1 gigabit per second, uses baseband transmission, NICs are installed in a network server, but only
and has a maximum cable length (from computer one is active. If the active NIC fails, the second
to hub) of 100 meters. NIC is automatically activated.
á 1000BASE-X (IEEE 802.3z) uses fiber-optic á Adapter load balancing is achieved when two or
cable, has a data-transfer rate of 1 gigabit per more NICs attached to the same network are
second and uses baseband transmission. The installed in a network server, all of them being
maximum cable length is 2 kilometers. utilized at the same time.
á 10 Gigabit Ethernet (802.3ae) uses fiber-optic á Port aggregation allows multiple NICs to act
cable exclusively to transfer data at 10 gigabits per together to increase the bandwidth available to
second. It uses only fiber-optic cable. It is antici- the network server. This requires an Ethernet
pated that the new standard will allow data trans- switch that supports port aggregation.
mission speeds of 10 gigabits per second for dis-
á NICs with Wake-On-LAN (WOL) technology
tances of 100 meters using multimode fiber-optic
allow the NIC to power up the computer in
cables and distances of at least 2 kilometers using
which it is installed upon receipt of a special
single-mode fiber-optic cable.
WOL magic packet.
á IBM created Token Ring network technology.
á A repeater is a network device that extends the
The original Token Ring network had a data-
length of a network by attaching multiple net-
transfer speed of 4Mbps and used shielded
work cable segments together and regenerating all
twisted-pair (STP) cable. Token Ring was stan-
data received from one cable segment and send-
dardized by IEEE as the 802.5 standard.
ing it to the other cable segment.
á Later versions of Token Ring had data-transfer
á A hub is a “repeater” with multiple ports.
speeds of 16Mbps, and could use either STP or
UTP cable.
15 Serv+ Ch_FF 7/10/01 10:02 AM Page 505
á One of the most popular protocols is TCP/IP QIC-80 MC-2120Extra 400MB/800MB Floppy,
(Transmission Control Protocol/Internet parallel
Protocol), mostly because it is the protocol used QIC-80XL MC-2120XL 170MB/340MB Floppy,
on the Internet. parallel
QIC-3020XL MC-3020XL 680MB/1.36GB Floppy,
á Another popular protocol is IPX/SPX used in parallel,
Novell NetWare networks. IDE
á NetBEUI is a protocol that is simple to imple- QIC-3020XL MC-3020Extra 1.6GB/3.2GB Floppy,
ment; however, it is not routable and therefore parallel,
IDE
cannot be used to build large networks.
QIC-3095 MC-3095 4GB/8GB IDE,
á The Simple Network Management Protocol SCSI-2
(SNMP) is a standard TCP/IP protocol that is
QIC-3220 MC-3220 10GB/20GB SCSI-2
used to manage network devices.
QIC-5010 DC-5010 16GB/32GB SCSI-2
á The Desktop Management Interface (DMI) is a
QIC-5210 DC-5210 25GB/50GB SCSI-2
means to collect information about a computer’s
operating environment and reporting it to a cen-
tral information point.
á Travan cartridge tapes are an enhancement to
á SAF-TE (SCSI Accessed Fault-Tolerant QIC technology with some backward compatibil-
Enclosures) is a methodology for the management ity with QIC tapes.
of external enclosures containing SCSI devices.
á Table 5 summarizes Travan tape drive standards.
á The most popular backup hardware is magnetic
tape drives.
á QIC (Quarter Inch Cartridge) tape drives are
generally found in entry-level servers due to their
relatively limited storage capacity.
á Table 4 summarizes QIC tape standards.
15 Serv+ Ch_FF 7/10/01 10:02 AM Page 506
Travan NS-20 NS-20 10GB/20GB SCSI-2, á Advanced Intelligent Tape (AIT) uses 8mm tape
EIDE technology but also contains Memory-In-Cassette
(MIC), which contains the tape log.
á Table 6 summarizes QIC and Travan compatibility. á Table 8 summarizes AIT tape technology.
á Table 9 summarizes DAT tape technology. á Table 11 summarizes LTO tape technology.
á High availability is the goal of keeping the net- á RAID 4, which requires a minimum of three
work server operational 24 hour per day, 7 days disk drives to implement, uses block-level parity
per week. This is accomplished by using methods to implement disk fault tolerance. The parity
such as redundancy, fault tolerance, RAID, UPS, information is stored on a single disk drive in
and clustering. the disk array.
á An uninterruptible power supply (UPS) is a very á RAID 5, which requires a minimum of three disk
necessary part of a network server. It provides drives to implement, uses block-level parity to
temporary power to the network server when the implement disk fault tolerance. The parity infor-
electrical power fails. mation is stored across all the disk drives in the
disk array.
á RAID (redundant array of inexpensive disks) is a
technique for providing fault tolerance in the á RAID 0/1 is a combination of RAID 0 (disk
event of a disk drive failure. striping without parity) and RAID 1 (disk mir-
roring or duplexing). This provides the fault tol-
á RAID 0 was not part of the original RAID speci-
erance of RAID 1 with the disk input and output
fication and does not provide any fault tolerance.
performance of RAID 0, but it requires a mini-
However, it does provide improved disk input
mum of four disk drives to implement.
and output performance.
á To implement RAID using hardware, a disk con-
á RAID 1 provides redundancy by copying data
troller that supports RAID is required. Most RAID
to two separate disk drives. In the event of one
disk controllers support the use of SCSI disk drives.
disk drive failure, the data is still available on the
second disk drive. á Some network operating systems support the
implementation of RAID solutions using soft-
á RAID 1 can be implemented in two different
ware. This requires the use of the network server
ways. When both disk drives are attached to the
processor to support the RAID solution.
same disk controller, the RAID solution is known
as disk mirroring. á A video adapter and monitor that supports a
video resolution of 1024 × 768 and 64K colors is
á When both of the RAID 1 disk drives are
adequate for most network server consoles.
attached to different disk controllers, the RAID
solution is known as disk duplexing. Disk á The Universal Serial Bus (USB) supports up to
duplexing eliminates the single point of failure 127 devices at a data-transfer rate of 12Mbps.
(the disk controller) that is the weak part of disk Uses of the USB port on a network server include
mirroring. UPS monitoring, and the attachment of such
devices as a mouse, a keyboard, or a modem.
á RAID 2, which requires a minimum of three disk
drives to implement, uses a hamming code to á IEEE 1394 (FireWire) is a serial implementation
implement disk fault tolerance. RAID 2 is not of SCSI-3. It supports up to 63 devices at a data-
used in any commercial RAID hardware. transfer rate of 400Mbps.
á RAID 3, which requires a minimum of three disk á Network servers usually include other devices,
drives to implement, uses bit-level parity to such as CD-ROM or DVD-ROM drives and a
implement disk fault tolerance. 3.5-inch floppy disk drive.
15 Serv+ Ch_FF 7/10/01 10:02 AM Page 509
á To minimize the space required by multiple á The installation, configuration, and maintenance
network servers, they can be mounted into a of the network server should be recorded (docu-
server rack. mented) in a logbook.
á Network servers and other components, such á A installation plan is a document that defines the
as a UPS, can be acquired in special rack-mount role of the network server along with its hardware
configurations. configuration.
á There are several network operating systems avail-
able for network servers, including the following:
• Novell NetWare 4 1.0—INSTALLATION
• Novell NetWare 5
This job dimension covers the pre-installation and instal-
• Microsoft Windows NT Server 4 lation of network server hardware, and it is the topic of
• Microsoft Windows 2000 Server 17% of the exam questions. The main points follow:
• Red Hat Linux á The server hardware specialist should verify that
the installation site for the network server has
• SuSE Linux adequate power (and that it is of the correct volt-
• VA Linux age and amperage) and cooling to support the
new network server.
• IBM OS/2 Warp Server
á The server hardware specialist should verify that
á Many network server hardware vendors supply there is a live network connection at the installa-
software to aid in the installation and configura- tion site for the new network server.
tion of the network operating system.
á The server hardware specialist should verify that
á Many network server hardware vendors supply the uninterruptible power supply (UPS) is of
hardware and software to aid in the management adequate size to support the new network server
of the network server hardware independent of and that the power plug on the UPS matches
the network operating system. the power outlet at the installation site.
á It is highly recommended that antivirus software
be installed on the network server.
15 Serv+ Ch_FF 7/10/01 10:02 AM Page 510
á If the network server is to be installed in a server á A network server might have more than one system
rack, the server hardware specialist should verify bus. Multiple buses can be either bridged or peer.
that all the network server components are
á Bridged buses share the bus bandwidth.
designed to fit into the rack.
á Peer buses provide additional bandwidth because
á The server hardware specialist should verify that
each can transfer data at full speed.
the processors to be installed in the network serv-
er are of the same speed, cache size, and stepping. á Adapters (disk controllers, NICs, and so on)
should be installed in such a manner as to
á The server hardware specialist should verify that
balance the load on peer buses.
the memory to be installed in the network server
is the correct size and amount and that it is com- á When two ATA (includes IDE, EIDE, ATA-2,
patible with the memory slots available in the Ultra ATA) are attached to the same ATA chan-
network server. nel, the disk drive role must be set to master on
one disk drive and slave on the other disk drive.
á The server hardware specialist should verify that
A third setting on some ATA disk drives, called
the disk drives and controllers are of the size
cable select (CSEL), automatically sets the role of
and type (ATA or SCSI) as specified by the
the disk drive.
installation plan.
á Each SCSI devices on a channel must have a
á The server hardware specialist should verify that
unique SCSI ID number, including the SCSI
the backup system is adequate to meet the needs
controller itself. A single SCSI channel can have
of the network server backup strategy.
up to 7 devices attached (15 devices on a Wide
á The server hardware specialist should verify that SCSI channel).
the SCSI cables are of the correct length and have
á Both ends of the SCSI channel must be terminat-
the correct number and type of connectors to
ed. There are three types of SCSI termination:
attach the SCSI devices to the SCSI controller.
passive, active, and forced perfect.
á The server hardware specialist should verify that
á SCSI has three different signaling systems: single
the disk controller type (ATA, SCSI-1, SCSI-2,
ended (SE), high-voltage differential (HVD), and
SCSI-3, FC-AL) and the number of channels and
low-voltage differential (LVD). The signaling
the bus type (PCI, EISA, PCI-X) is as specified in
system dictates the length of the SCSI channel.
the installation plan.
á There are three different types of SCSI connec-
á The server hardware specialist should verify that the
tors: 50-pin, 68-pin, and 80-pin (SCA).
network interface cards (NICs) are of the correct
type (Ethernet, Fast Ethernet, Gigabit Ethernet, á The 50-pin connectors (both internal and external
Token Ring) as specified in the installation plan. versions) are used by Narrow (8-bit) SCSI channels.
á The server hardware specialist should follow á The 68-pin connectors (both internal and external
ESD (electrostatic discharge) best practices versions) are used by Wide (16-bit) SCSI channels.
during the installation of the network server á The 80-pin SCA connector is a special-use
and its components to prevent damage to them connector that not only allows the transfer of
that can be caused by the buildup and discharge data, but also supplies power to the SCSI device
of static electricity. and SCSI ID configuration.
15 Serv+ Ch_FF 7/10/01 10:02 AM Page 511
á When connecting an Ethernet NIC to an á A server hardware specialist should verify that the
Ethernet hub, a standard Category 5 unshielded system BIOS on the network server is the latest
twisted-pair (UTP) cable that contains 4-pair available by checking the network server vendor
(8 wires) is used. The patch cable is terminated web site. If the current system BIOS is not the
with an RJ-45 connector. latest available, download and install the latest
system BIOS.
á A special patch cable that attaches the transmit
pair to the receive pair on the other end of the á Other network server hardware components,
cable is called a cross-over cable, and it is used such as the SCSI controller and the RAID
to attach two hubs together or to attach two controller, also have a BIOS. These can also be
computers together. upgraded, but only do so when absolutely neces-
sary to fix a problem.
á When installing a network server and its compo-
nents into a server rack, the heavier components á Before upgrading the BIOS on a SCSI or RAID
(usually the UPS) should be installed near the controller, always back up the data on the net-
bottom of the rack. work server’s disk drives.
á External devices attached to the network server á A hardware RAID system must be configured
should be powered up before the network server using software provided by the RAID system
is powered up. manufacturer before the network operating
system can be installed.
á During power-up of the network server, the num-
ber of processors installed in the network server, á The RAID system configuration software allows
the amount of memory in the network server, and the server hardware specialist to select which of
any devices attached to a SCSI controller should the disk drives that are connected to the RAID
be displayed on the network server monitor. controller are to be combined into a RAID con-
figuration and which RAID configuration that
á Always document the network server installation
they will be combined into.
in the server logbook.
á The disk overhead for implementing RAID 1
is 50%. For example, two 18GB disk drives
configured as RAID 1 will provide 18GB of
2.0—CONFIGURATION storage capacity.
á The disk overhead for implemented RAID 5 is
This job dimension covers the configuration of network
((1/n) * 100)%. For example, five 18GB disk
server hardware and software, and it is the topic of
drives configured as RAID 5 will have a disk
approximately 18% of the exam questions. The main
overhead of ((1/5) * 100)% or 20%. The total
points follow:
storage capacity will be 72GB.
á Table 12 summarizes RAID Characteristics.
15 Serv+ Ch_FF 7/10/01 10:02 AM Page 512
á Server log files often need to be configured to á Always review the instructions and Frequently
specify a maximum log file size and the action to Asked Questions (FAQs) before installing a soft-
take when the log file is full. ware upgrade.
á A swap file is used by network server operating á Test the upgrade on a non-production system
systems that support virtual memory. before it is installed on a production system.
á The swap file for Microsoft Windows NT Server á Time must be scheduled to take the network server
4 and Microsoft Windows 2000 Server is out of production while the upgrade is installed.
C:\PAGEFILE.SYS .
á Always perform a full system backup before
á The swap file for IBM OS/2 Warp Server is installing the upgrade.
C:\OS2\SYSTEM\SWAPPER.DAT.
á Implement electrostatic discharge best practices
á The swap file should be located on a disk drive while performing the upgrade.
that does not contain other frequently accessed
á After installing the upgrade, verify that it is rec-
files.
ognized by the network server and the network
á The four major components of a network server operating system.
that should be monitored for performance are
á Perform a new server baseline after the upgrade
processor(s), memory, the disk subsystem, and
has been performed.
the network subsystem.
á Document the upgrade in the network server
á The network server configuration should be
logbook.
documented in the server logbook.
á One way to upgrade the processor in a network
server is to replace the existing processor with a
faster processor.
3.0—UPGRADING á Some network servers have the capability to
install multiple processors. In this case, installing
This job dimension covers the upgrading of network
additional processors is a way to upgrade the
server hardware and software and is the topic of
network server.
approximately 12% of the exam questions. The main
points follow: á Additional processors in a network server must
match the existing processors in speed, amount
á When a server hardware specialist performs an
of Level 2 cache memory, and stepping (within
upgrade to the network server, the upgrade
1 number of the other processors).
checklist should be followed.
á Upgrading to a faster processor or installing an
á The latest drivers for a network server component
additional processor may require an upgrade to
can be found at the hardware component ven-
the system BIOS.
dor’s web site.
á Some configuration changes might be required in
á The latest patches for the network server operat-
the network operating system when upgrading
ing system can be found at the operating system
from one processor to multiple processors.
vendor’s web site.
15 Serv+ Ch_FF 7/10/01 10:02 AM Page 514
á When adding a second disk drive to an ATA á An differential backup copies all the files on the
channel, one disk drive must be configured in the network server that have changed or been created
master role and the other disk drive must be con- since the last full backup to a backup device.
figured in the slave role.
á Table 13 summarizes the differences among these
á When adding a disk drive to a SCSI channel, it three backup types.
must be configured with a unique SCSI ID.
á Before adding memory to the network server, ver-
ify that the network server and network operating SUMMARY TABLE 13
system can recognize the additional memory. BACKUP TYPES
Type Includes Only Turns Archive Typical
á Additional memory must match the existing Files with Bit Off upon Running
memory in type (EDO, SDRAM, RDRAM), Archive Bit On Completion Time
form factor (SIMM, DIMM, RIMM), speed, and (1=Longest)
error correcting capability (ECC, non-ECC). Full No Yes 1
á The system BIOS might need to be upgraded Incremental Yes Yes 2
to support larger memory modules, larger disk
Differential Yes No 3
drives, or faster processors, or to correct errors
in the existing BIOS.
á PCI hot plug allows an adapter to be replaced, á Don’t mix incremental and differential backups.
upgraded, or added without powering down the á One set of backup tapes should be stored off site.
network server.
á Microsoft Windows NT Server 4 and Microsoft
Windows 2000 Server include a Backup utility.
á UNIX/Linux system use tar and cpio to perform
4.0—PROACTIVE backups.
MAINTENANCE á When changes are made to the hardware or soft-
ware on a network server, the server hardware
This job dimension covers proactive maintenance, specialist should perform a new set of baseline
which can be summed up as everything you do to a net- measurements.
work to ensure the system will continue to perform in
á The Simple Network Management Protocol
the future. This job dimension comprises approximately
(SNMP) is a standard part of the TCP/IP
9% of the exam questions. The main points follow:
protocol suite designed to allow management
á A full backup copies all the files on the network of network devices.
server to a backup system (usually a magnetic
á The SNMP agent is a small program that runs on
tape drive).
the device to be monitored.
á An incremental backup copies all the files on the
á The SNMP management station queries the
network server that have changed or been created
agents on the network devices for information
since the last backup to a backup device.
about the device.
15 Serv+ Ch_FF 7/10/01 10:02 AM Page 515
á The SNMP agent uses a Management á The temperature of the server room should be
Information Base (MIB), which defines the items below the maximum operating temperature speci-
to be monitored on the network device. fied for the network server.
á A SNMP trap is a condition, which is set on the á The humidity of the server room should be with-
SNMP agent, which is reported to the SNMP in the range specified for the network server.
management station when the condition is met.
á The server room should be protected from
á An SNMP community name is a primitive secu- electrostatic discharge (ESD). Such things as
rity system that defines which SNMP agents can antistatic carpet help reduce ESD.
communicate with which SNMP management
á An uninterruptible power supply (UPS) should
stations.
be installed to provide power to the network
á A network server should be installed in an envi- server during electrical power failures, as well
ronmentally appropriate controlled room (tem- as to provide protection to the network server
perature, humidity, dust free). from power surges.
á Verification of the hardware in a network server á The server room should be located in a part of
can be done using software that comes with the the building that is unlikely to flood.
network server operating system.
á There should be a fire suppression system in the
á Some hardware verification can also be done server room.
using the system BIOS.
á A backup generator will keep the network server
á Remote notification of problems can be done operational for long periods of electrical outage.
using SNMP or utilities that are available with
some network server operating systems.
6.0—TROUBLESHOOTING AND
5.0—ENVIRONMENT PROBLEM DETERMINATION
This job dimension requires an understanding of the This job dimension covers the troubleshooting and prob-
physical environment of the network server and is the lem determination of network server hardware and soft-
topic of approximately 5% of the exam questions. The ware. This job dimension is the topic of approximately
main points follow: 27% of the exam questions. The main points follow:
á A network server should be installed in a secure á The same six questions used by a good reporter
room. (who, what, where, when, why, and how), can be
used by a server hardware specialist to determine
á Make sure that the backup tapes, which contain the cause of a problem.
copies of all the data on the network server, are
also stored in a secure location. á A server hardware specialist should have a list of
contacts that are available to help with specific
á If the network server is mounted in a server rack, problems (for example, the building superintendent
make sure that the service rack doors are locked. for electrical power or air-conditioning problems).
15 Serv+ Ch_FF 7/10/01 10:02 AM Page 516
á A server hardware specialist should be able to use á To shut down a Novell NetWare server, issue the
his or her senses to help find problems or condi- down command from the system console.
tions that can contribute to problems.
á To shut down UNIX/Linux server use the
á The primary system diagnostic tools available in shutdown command.
Microsoft Windows NT Server are Task Manager,
á To shut down an IBM OS/2 Warp Server, select
Windows NT Diagnostics, and the Event Viewer.
the Shutdown button on the desktop.
á The primary network diagnostic tools available in
á Field replaceable unit (FRUs) are network server
Microsoft Windows NT Server are, ipconfig
components that can be easily replaced on site.
nbtstat, netstat, nslookup, ping, and tracert.
á Event logs are files that contain information on
á The primary system diagnostic tools available in
events that have occurred on the network server.
Microsoft Windows 2000 Server are Task
These are usually maintained by the network
Manager, System Information, and Event Viewer.
server operating system.
á The primary network diagnostic tools available in
á A health log records the failure of network server
Microsoft Windows 2000 Server are ipconfig,
hardware components that is usually maintained
nbtstat, netdiat, netstat, nslookup, ping, tracert,
by special software supplied by the network server
and pathping.
hardware vendor.
á The primary diagnostic tool available in Novell
á When troubleshooting a failure of a network
NetWare is Monitor.
server, be sure to check the server logbook for
á The primary network diagnostic tool available in recent changes in the hardware and software on
Novell NetWare is Config. the network server.
á The primary diagnostic tools available in á When troubleshooting a network server failure,
UNIX/Linux are iostat, vmstat, ps, and sar. be sure to check the network server hardware
vendor’s web site, the network operating
á The primary network diagnostic tools available
system vendor’s web site, and related Internet
in UNIX/Linux are ifconfig, netstat, ping, and
newgroups for information on similar failures.
traceroute.
á If it appears that the network server failure is
á Before shutting down a network server, it is good
beyond your abilities to solve in a timely manner,
practice to send a message to all the users logged
call a more experienced server hardware specialist
in to the network server stating that the network
to assist in solving the problem.
server is going to be shut down.
á Wake-On-LAN technology can be used to
á To shutdown Microsoft Windows NT Server and
power up a computer attached to a network for
Microsoft Windows 2000 Server, press the Start
troubleshooting or backup purposes.
button and select Shutdown.
15 Serv+ Ch_FF 7/10/01 10:02 AM Page 517
á A remote alert is generated by SNMP when a á In the event that a network server fails to boot
preset condition (trap) occurs. up, insert a POST Diagnostics Board in an
empty adapter slot and read the error code form
á Four major components can cause a bottleneck in
the POST Diagnostics Board displays to deter-
the network server: the processor, the memory,
mine the cause of the boot-up failure.
the disk subsystem, and the network subsystem.
á Most major network operating systems provide
software that can be used to measure the utiliza-
tion and performance of the major components 7.0—DISASTER RECOVERY
in the network server.
This job dimension deals with the planning and imple-
á A processor utilization percentage that is con-
mentation of a disaster-recovery process and is the topic
stantly greater than 70% indicates that the
of 12% of the exam questions. The main points follow:
processor is the system bottleneck.
á Disaster recovery is a set of activities that are
á A processor bottleneck can be eliminated by
designed to reduce the likelihood, duration, and
replacing the existing processor with a faster
impact of a disaster.
processor, or by installing additional processors in
a multiprocessor-capable system. á To reduce the likelihood of a disaster striking a
network server, use redundant hardware in the
á A disk subsystem utilization percentage that is
network server.
constantly greater than 90% indicates that the
disk subsystem is the system bottleneck. á Redundancy for hard disk drives is accomplished
by implementing RAID (redundant array of inex-
á Lack of memory in a network server that sup-
pensive disks).
ports virtual memory will cause many pages to be
written to disk, thus increasing the utilization of á RAID 1 (mirroring or duplexing) provides
the disk subsystem. redundancy by duplicating all data on pairs
of disk drives.
á Installing additional memory into the network
server can eliminate a memory bottleneck. á RAID 5 (striped set with parity) provides redun-
dancy by calculating parity information for each
á Computer viruses can manifest themselves as
block of data written to the disks in the array.
either hardware or software problems. Always
update the virus signature file and run the virus- á RAID 0/1 provides redundancy by mirroring a
scan software as a standard part of troubleshoot- striped set without parity onto another striped set
ing a network server problem. without parity.
15 Serv+ Ch_FF 7/10/01 10:02 AM Page 518
á A hot spare disk drive on a RAID implementa- á A disaster-recovery plan consists of the server
tion allows the RAID system to “rebuild” the hardware specialist’s risk analysis and suggestions
RAID disk set after a single drive failure by using for using fault-tolerant methods to minimize
the “spare” drive for writing data as soon as the these risks. The disaster-recovery plan also
single drive failure is detected. includes a list of spare parts to be kept on hand
in the vent of a component failure.
á Other network server components that are often
configured in a redundant manner are network á A big part of disaster recovery is having copies of
server power supplies, cooling fans, network the data that is stored on the network server
interface cards, and processors. available on some sort of backup media (usually a
magnetic tape).
á An uninterruptible power supply (UPS) provides
power to the network server components in the á Various tape storage devices are available for net-
event of the electrical power failure. A UPS is work servers, including Digital Audio Tape
designed to provide electrical power for a very (DAT) (using Digital Data Storage, or DDS),
short time. If a longer period of time is required, Digital Linear Tape (DLT), Mammoth tape,
an emergency electric generator is required. Linear Tape-Open (LTO), and Advanced
Intelligent Tape (AIT).
á A server cluster is a group of network servers
working together as a single system. A loss of a á Tape backup schemes include full backup, incre-
single network server in the cluster does not stop mental backup, and differential backup.
mission-critical applications from running,
á A copy of the backup tapes should be stored off
because the other network servers in the cluster
site (in a different location from the network
remain operational. A server cluster can be con-
server). After all, if the backup tapes are stored in
sidered to be “redundant servers.”
the same room or in the same building as the
á A hot swap component, such as a disk drive or network server, a building fire puts you out of
power supply, can actually be replaced while the business.
network server is fully operational, resulting in
á The disaster-recovery plan should be written
zero downtime.
down and it should be reviewed regularly and
á A hot spare is a disk drive that is powered up but updated as needed.
that is not being used; however, it is available to
á A hot site is a commercial facility that is available
automatically take the place of a failed drive in a
to provide a fully operational system once your
hardware RAID array.
data is loaded onto their network server.
á A warm swap is a compromise between hot swap
á A cold site is an empty room that your existing
and hot spare techniques.
network servers can be relocated into if the pri-
mary site is destroyed or becomes unavailable.
16 Serv+ Ch_Tips 7/11/01 10:06 AM Page 519
LEARNING AS A PROCESS
To better understand the nature of preparation for the
exams, it is important to understand learning as a
process. You probably are aware of how you best learn
new material. You may find that outlining works best
for you, or you may need to “see” things as a visual
learner. Whatever your learning style, test preparation
takes place over time. Obviously, you cannot start
studying for this exam the night before you take it; it is
very important to understand that learning is a devel-
opmental process. And as part of that process, you
need to focus on what you know and what you have
yet to learn.
Learning takes place when we match new information
to old. You have some previous experience with com-
puters, and now you are preparing for this certification
exam. Using this book, software, and supplementary
materials will not just add incrementally to what you
know; as you study, you will actually change the
organization of your knowledge as you integrate this
new information into your existing knowledge base. This
will lead you to a more comprehensive understanding of
the tasks and concepts outlined in the objectives and of
computing in general. Again, this happens as a repeti- Study and Exam
tive process rather than a singular event. Keep this
model of learning in mind as you prepare for the exam,
and you will make better decisions concerning what to
Prep Tips
study and how much more studying you need to do.
16 Serv+ Ch_Tips 7/10/01 10:03 AM Page 520
Common Sense Strategies You should set a goal for your pre-testing. A reasonable
Finally, you also should follow commonsense practices goal would be to score consistently in the 90% range.
when studying. Study when you are alert, reduce or See Appendix D, “Using the ExamGear, Training
eliminate distractions, take breaks when you become Guide Edition Software,” for a more detailed explana-
fatigued, and so on. tion of the test engine.
For most people, this can be difficult to assess objec- An original fixed-form, computerized exam is based on
tively on their own. Practice tests are useful in that they a fixed set of exam questions. The individual questions
reveal more objectively what you have learned and are presented in random order during a test session. If
what you have not learned. You should use this infor- you take the same exam more than once, you will see
mation to guide review and further study. the same number of questions, but you won’t necessari-
Developmental learning takes place as you cycle ly see the exact same questions. This is because two or
through studying, assessing how well you have learned, three final forms are typically assembled for such
reviewing, and assessing again until you think you are exams. These are usually labeled Forms A, B, and C.
ready to take the exam. As suggested previously, the final forms of a fixed-form
You may have noticed the practice exam included in this exam are identical in terms of content coverage, num-
book. Use it as part of the learning process. The ber of questions, and allotted time, but the questions
ExamGear software on the CD-ROM also provides a differ. You may notice, however, that some of the same
variety of ways to test yourself before you take the actual questions appear on, or rather are shared among,
exam. By using the practice exam, you can take an entire different final forms. When questions are shared among
timed, practice test quite similar in nature to that of the multiple final forms of an exam, the percentage of sharing
actual Server+ exam. The ExamGear Adaptive Exam is generally small. Many final forms share no questions,
option can be used to take the same test in an adaptive but some older exams may have a 10% to 15%
testing environment. This mode monitors your progress duplication of exam questions on the final exam forms.
as you are taking the test to offer you more difficult Fixed-form exams also have a fixed time limit in which
questions as you succeed. By using the Study Mode you must complete the exam. The ExamGear test
option, you can set your own time limit, focus only on a engine on the CD-ROM that accompanies this book
particular domain (for instance, Configuration) and also provides fixed-form exams.
receive instant feedback on your answers.
16 Serv+ Ch_Tips 7/10/01 10:03 AM Page 522
Finally, the score you achieve on a fixed-form exam is Putting It All Together
based on the number of questions you answer correctly.
The exam’s passing score is the same for all final forms Given all these different pieces of information, the task
of a given fixed-form exam. now is to assemble a set of tips that will help you suc-
cessfully tackle the Server+ certification exam.
Table 1 shows the format for the exam.
á Carefully read all information and instruction confidently, noting the relative difficulty of each
screens. These displays have been put together to question, on the scratch paper provided. After
give you information relevant to the exam you you have made it to the end of the exam, return
are taking. to the more difficult questions.
á Read the exam questions carefully. Reread each á If session time remains after you have completed
question to identify all relevant details. all questions (and if you aren’t too fatigued!),
review your answers. Pay particular attention to
á Tackle the questions in the order they are pre-
questions that seem to have a lot of detail or that
sented. Skipping around will not build your con-
involved graphics.
fidence; the clock is always counting down.
á As for changing your answers, the general rule of
á Do not rush, but also do not linger on difficult
thumb here is don’t! If you read the question care-
questions. The questions vary in degree of diffi-
fully and completely and you thought like you
culty. Don’t let yourself be flustered by a particu-
knew the right answer, you probably did. Do not
larly difficult or verbose question.
second-guess yourself. If as you check your
á Note the time allotted and the number of ques- answers, one clearly stands out as incorrectly
tions appearing on the exam you are taking. marked, of course you should change it. If you
Make a rough calculation of how many minutes are at all unsure, however, go with your first
you can spend on each question and use this to impression.
pace yourself through the exam.
If you have done your studying and follow the preced-
á Take advantage of the fact that you can return to ing suggestions, you should do well. Good luck!
and review skipped or previously answered ques-
tions. Record the questions you cannot answer
16 Serv+ Ch_Tips 7/10/01 10:03 AM Page 524
17 Serv+ Ch_PE 7/11/01 10:09 AM Page 525
Practice Exam
17 Serv+ Ch_PE 7/10/01 10:04 AM Page 526
16. Which of the following do not require drivers? 20. What can be used to ascertain the performance of
various network server components under normal
A. SCSI controllers
operating conditions?
B. RAID controllers
A. A performance-monitoring tool
C. Sound cards
B. An antivirus program
D. Email accounts
C. A workload indexer
17. Where should you install an SNMP agent?
D. Both A and C
i.On the network routers
21. What is the purpose of an SNMP?
ii.On a peripheral network
A. To manage user privileges
iii. On your network operating system
B. To manage diagnostic software
iv. On a standalone machine
C. To manage network devices
A. iv only
D. To manage network capacity
B. i, ii, and iv
22. Which of the following is not a server-monitor-
C. i and iii ing agent?
D. ii only A. Compaq Insight Manager
18. After you have installed an SNMP agent, you B. Dell OpenManage IT Assistant
must configure it with which name?
C. HP Toptools
A. An SNMP community name
D. Java Servlet Engine
B. An SNMP username
23. During the verification of a backup, to what do
C. An SNMP network name you compare your backed-up media?
D. A URL address A. The files you attempted to back up
19. Which term refers to a measurement of how the B. The CD-ROM content
server is currently operating and which can be
C. The access control list
compared against at a later time?
D. The tape vendor’s web site
A. Server baseline
24. When should you perform a full backup of the
B. Server standards
entire contents of a network server?
C. Server diagnostics
A. Before running the POST tests
D. Server partitions
B. Before attempting any upgrade operation
17 Serv+ Ch_PE 7/10/01 10:04 AM Page 529
C. Before logging off the system 28. What are disk drive speeds characterized by?
D. Daily A. Rotational speed
25. To add another processor to a multiprocessor- B. Speed at which they access the RAM cache
capable network server, the new processor must
C. Different factors by different manufacturers
meet which of the following criteria?
D. Bidirectional speed
i.Be the same model processor
29. Why should you avoid installing a memory mod-
ii.Have the same clock speed
ule with tin leads into a memory slot with gold
iii. Must match the stepping within 1 version connectors?
iv. Must have the same Level 2 (L2) cache size A. The combination can increase electrostatic
discharge.
A. i and ii
B. The combination is cost prohibitive for all
B. i and iii
but the largest of organizations.
C. ii and iv
C. The combination can slow down application
D. i, ii, iii, iv interoperability.
26. If your original installation of Microsoft D. The combination can cause corrosion.
Windows 2000 Server was on a network server
30. What is the most reliable way to check the cur-
with a single processor, what must you update on
rent memory configuration on a network server?
the network server if you now want to use multi-
(Choose the best answer.)
ple processors?
A. Call the vendor’s help desk.
A. The per-seat license
B. Open the server’s chassis and check.
B. The Add New Hardware icon
C. Check the memory vendor’s web site.
C. The hardware abstraction layer
D. Open multiple, memory-intensive programs.
D. The Multitasking dialog box
31. How can you determine the version of the BIOS
27. When adding additional disk drives, what is the
currently in a network server?
best way to ensure interoperability between the
old and new drives? A. The BIOS brand and version number usually
appear on a sticker affixed to the back of the
A. Check the new drive’s dynamic link library.
network chassis.
B. Acquire disk drives from the same
B. The BIOS brand and version number usually
manufacturer.
display when you attempt to open a corrupt-
C. Read the Help files. ed file.
D. Check the IEEE 802 standard. C. The BIOS brand and version number usually
do not display on a server. To obtain this
information, call the vendor’s help desk.
17 Serv+ Ch_PE 7/10/01 10:04 AM Page 530
D. The BIOS brand and version number usually B. A Directory Memory Access channel for
display on the system monitor when the net- the adapter
work server is powered up.
C. A USRT
32. During a flash BIOS upgrade, when is the old
D. A VLSI
BIOS erased?
36. To what are external devices considered external
A. It is not erased; it resides in the EEPROM.
of?
B. After the new BIOS code is written to the
A. The printer
EEPROM.
B. The monitor
C. Before the new BIOS code is written to the
EEPROM. C. The server chassis
D. While the new BIOS code is being written D. The modem
to the EEPROM. 37. How do system-monitoring agents report infor-
33. With regard to PCI hot plug, what is hot mation to a management console?
expansion? A. A NIC
A. Removing a failed PCI adapter and inserting B. The IRQ channel
an identical adapter into the same slot while
the network server is operational C. Standard protocols such as HTTP, SNMP,
and DMI
B. Replacing an existing adapter with an upgrad-
ed adapter while the network server is running D. The error log
C. Installing a new adapter into a previously 38. What can system-monitoring agents sometimes
empty slot while the network server is running predict?
D. Closing the slot release lever to secure an A. Impending upgrades to system software
adapter in the PCI slot B. Impending failure of network server
34. What does a RAID controller use onboard components
memory for? C. Impending extensions required by hardware
A. A networking interface updates
B. A buffer D. Impending browser wars
C. A firmware extension 39. Which of the following is/are general categories
of service tools?
D. A RIMM update
A. Diagnostic tools
35. If a peripheral’s adapters are not Plug and Play,
you might have to configure what? B. Diagnostic partition
A. A flow control port C. Server support utilities
D. All of the above
17 Serv+ Ch_PE 7/10/01 10:04 AM Page 531
40. What sort of service tools enable you to view and A. Incremental
change the configuration of the hardware compo-
B. Optional
nents in the network server without the network
server operating system being loaded? C. Full
A. Diagnostic partition utilities D. Differential
B. Server support utilities 45. Which utility do you use to monitor system
performance in Windows 2000?
C. EISA configuration utilities
A. Performance Tool
D. Diagnostic tools
B. IIS CPU Throttling
41. If a UPS replacement requires that the original
UPS be turned off before replacement, what else C. Event Viewer
must you do? D. Resource Meter
A. Turn the modem off. 46. A baseline enables you to do which of the
B. Shut down the network server. following?
C. Reboot the network server. A. View shared objects on the network
D. Call the vendor’s help desk. B. Troubleshoot memory and processor problems
42. What questions should you answer before C. Set counters in Performance Tool
installing a new UPS? D. Compare system performance and note
A. Is the old power source adequate? deviations when changes are made
B. Is it the correct amperage? 47. At what port does the SNMP service run
by default?
C. Does the new UPS use the same type of
power plug that the old UPS is using? A. 80
D. All of the above B. 110
43. Which of the following most closely defines a C. 161
“regular” backup schedule? D. 21
A. Every week 48. Which of the following describes the get-next
B. Every day command in SNMP?
C. Depends on the value of your data and type A. Requests a specific value
of your business B. Changes the value on an object that has the
D. Every month properties of read-write
44. Which of the following is not a normal C. Requests the next object’s value
backup category? D. Run in Verbose mode
17 Serv+ Ch_PE 7/10/01 10:04 AM Page 532
49. What is a plenum? 54. In Windows 2000, one task of the Messenger
service is to what?
A. The outside covering of coaxial cable
A. Send alerts
B. The area around the server
B. Store hardware information
C. The space above a drop ceiling
C. Keep backup information current
D. EMI
D. Recover data
50. What kind of coaxial cable is prohibited from
existing in the plenum because of fire codes? 55. In a large environment with many people
requiring server access, which is the least
A. All
effective way to secure a server in a server
B. PVC room or computer data center?
C. Plenum-grade A. Combination locks
D. EMI-protected B. Proximity readers
51. What tool in Windows 2000 Server enables you C. Lock and key
verify COM ports?
D. Card readers
A. System Properties
56. Which of the following are built-in server
B. Device Manager locking mechanisms?
C. SNMP A. Locking cases and drives
D. Performance Tool B. Locking network cards
52. What utility in Linux provides system C. Locking motherboards
information?
D. Locking power sources
A. cpio
57. Which of the following can be used to alleviate
B. tar power-line issues?
C. proc i.UPS
D. tar cvf ii.Surge suppressor
53. What is the single best way to be notified of a iii. Antistatic devices
remote situation on a network?
iv. Clamping voltage
A. NTHQ
A. i and ii
B. SNMP
B. i, iii, and iv
C. FTP
C. ii only
D. PVC
D. ii, iii, and iv
17 Serv+ Ch_PE 7/10/01 10:04 AM Page 533
58. Which of the following environmental factors 61. What command can be used on a Novell server
should also be considered when designing and to see the IP stack configuration?
deploying a safe server environment?
A. ifconfig
i.Fire suppression
B. config
ii.Sound suppression
C. ipconfig
iii. ESD
D. winconfig
iv. Air quality
62. You need to reboot your Linux server and want
A. i and ii to do so as expeditiously as possible. What run
level can be used to reboot the system?
B. i, iii, and iv
A. 0
C. ii only
B. 1
D. ii, iii, and iv
C. 3
59. Your network consists of three 10BASE-2 subnets
with 25 nodes on each. Users on one subnet sud- D. 6
denly report that they are having difficulty com-
63. Users are complaining that they are having diffi-
municating with the network, and there are a
culty running applications from the terminal
high number of collisions. What would be the
server. Which item is most likely the bottleneck?
most likely cause of this?
A. Processor
A. Too many users on the node
B. RAM
B. Bad hub
C. Network I/O
C. Faulty terminator
D. Disk I/O
D. Malfunctioning transceiver
64. Users are receiving error messages that they are
60. Which of the following could be symptoms of an
unable to print after the accounting department
unhooked 10BASE-2 cable on a user’s machine?
submits large print jobs to the server. What would
i.Entire segment unable to communicate. be the most important element you could add to
the server to help alleviate this problem?
ii.User has an IP address of 169.254.2.33.
A. Faster network card
iii. User cannot print to network printer.
B. Faster processor
iv. Computer keeps rebooting.
C. Larger hard drive
A. iv only
D. More RAM
B. i, ii, and iv
65. You are the administrator for a small peer-to-peer
C. i, ii, and iii
network of Windows clients. You add a new
D. ii only computer to the network, but it cannot see other
resources. What would be the most likely cause?
17 Serv+ Ch_PE 7/10/01 10:04 AM Page 534
80. Which of the following steps should the server 3. B. When installing a rack-mounted server, you
hardware specialist perform first before beginning must follow the server vendor’s installation
the installation? instructions, have all rack-mount components,
and install special supports for each component
A. Verify that the server rack is assembled.
(install rails on all rack mount components).
B. Verify that the location of the server is close Answer A is incorrect as it is just one step that
to the server hardware specialist. must be followed. Answer C is incorrect as placing
C. Verify that there is another computer system the equipment in the rack according to size is
available for testing. inaccurate. Answer D is incorrect as it is just one
step that must be followed. See Chapter 2, section
D. Verify that everything specified in the installa- “Installing Hardware.”
tion plan is ready and available.
4. D. All externally powered devices should be
turned on first. Answer A is incorrect as it is the
second event that should occur. Answers B and C
ANSWERS AND EXPLANATIONS are incorrect as they occur later in the sequence.
See Chapter 2, section “Installing Hardware.”
Each of the answers below includes several things. You
are presented with the correct answer, then an explana- 5. A. When the network server goes through power
tion of why the correct answer is correct. This is followed up, the version number of the system BIOS
by an explanation of why the incorrect answers are usually displays. Answer B is incorrect because
incorrect. Finally, you are also given a cross-reference to the BIOS does not display upon logon. Answer
the chapter number and section where the information C is incorrect because the administrator cannot
pertaining to the question is covered. schedule the BIOS display. Answer D is incorrect
because the BIOS does usually display upon
1. D. The location of the server, the power outlet power up. See Chapter 3, section “Network
for the server, and the network connection for the Server Configuration.”
server should be included in an installation plan.
Answer ii is incorrect as a complete list of the 6. B. Most network servers have a BIOS that is
hardware should also be included in the plan. See flashable, meaning that it can be easily erased and
Chapter 2, section “Planning for Installation.” updated via software. Answer A is incorrect
because all BIOS is encoded in ROM, but not all
2. B. Hardware components of the server are dictat- BIOS can be updated via software. Answer C is
ed by the server’s role in the network. Answers A, incorrect because the BIOS is encoded in ROM,
C, and D are incorrect because even though cost, not RAM. Answer D is a false statement. See
expertise, and space may indeed be taken into Chapter 3, section “Network Server
consideration, it is the server’s role in the network Configuration.”
that commonly dictates the components used. See
Chapter 2, section “Planning for Installation.” 7. A. RAID 1 has two different implementations,
disk mirroring and disk duplexing. Answers B, C,
and D are incorrect because those RAID levels
17 Serv+ Ch_PE 7/10/01 10:04 AM Page 537
utilize disk striping or other implementations 12. A. Connecting the external disk subsystem to the
other than mirroring and duplexing. See Chapter network server might be accomplished by using a
3, section “Configure Hardware-based RAID.” standard external SCSI cable or a Fibre Channel
might be utilized. Answers B, C, and D are incor-
8. D. RAID 5 requires a minimum of three disk
rect because these are network protocols, not
drives to implement. Answers A, B, and C are
ways to connect the disk subsystem to the net-
incorrect because RAID 5 protects against a
work. See Chapter 3, section “Configuring
single disk drive’s failure causing the network serv-
External Peripherals.”
er to fail by enabling you to re-create the informa-
tion on the failed drive on-the-fly using the infor- 13. D. RTM is a generic term used by many software
mation on the remaining disks. Therefore, at least vendors as an acronym for “Release To
two other drives (besides the failed one) are Manufacturing”; always in relation to new prod-
required (for a total of 3). See Chapter 3, section ucts and not upgrades. Periodically network oper-
“Configure Hardware-based RAID.” ating system vendors issue updates to their net-
work operating systems. These updates have vari-
9. B. To test network connectivity when using the
ous names. (Microsoft calls them service packs;
TCP/IP protocol, all network operating systems
IBM calls them fixpacs; and Novell calls them
use the Ping command. Answers A, C, and D
patches.) See Chapter 3, section “Installing
are incorrect because they do not enable you to
Network Operation System Updates.”
check network connectivity to the Internet. See
Chapter 3, section “Installing the Network 14. D. Operating system updates usually fix bugs or
Operation System.” close security holes that have been found in the
released version of the operating system. Answer i
10. D. The ipconfig command is used to view
is incorrect because installation updates are not
or change network configuration in Windows
usually incorporated into operating system
NT/2000 systems. The inetcfg command is used
updates. Answer iv is incorrect as this is some-
to view or change network configuration in
thing that is addressed in major software revi-
Novell NetWare 4.x/5.x systems. The ifconfig
sions, not simple updates. See Chapter 3, section
command is used to view or change network con-
“Installing Network Operation System Updates.”
figuration in UNIX/Linux systems. Answers A, B,
and C are incorrect because they are incorporated 15. C. When updating manufacturer-specific drivers,
into choice D. See Chapter 3, section “Installing always check the version number of drives for
the Network Operation System.” devices in the network server. Answers A, B, and
D are incorrect answers because you first need to
11. C. On the high end an external disk subsystem
check whether you have the latest driver (which
chassis might have 100 or more disk drives in it.
you do by comparing the version numbers on
Answers A and B are incorrect because the high-
your devices against those for the latest drivers).
end system might have 100 or more drives.
See Chapter 3, section “Configuring Manufacturer-
Answer D is a false statement. See Chapter 3,
Specific Drivers.”
section “Configuring External Peripherals.”
17 Serv+ Ch_PE 7/10/01 10:04 AM Page 538
16. D. Email accounts do not require drivers because installed on them. SNMP was designed to make
they are not hardware devices attached to your it possible to manage network devices, such as
system; they are software applications. Answers A, routers. However, the use of SNMP has been
B, and C are incorrect because these are all hard- expanded to allow it to be able to manage practi-
ware devices that may be attached to your net- cally any device, including a computer, attached
work server and therefore require drivers to func- to a network. Answers A, B, and D are incorrect
tion with your server. See Chapter 3, section because SNMP is not used to manage these func-
“Configuring Manufacturer-Specific Drivers.” tions. See Chapter 3, section “Documenting the
Network Server Configuration.”
17. C. You should install the SNMP agent on your
network operating system. You can also install it 22. D. Java Servlet Engine is not a server-monitoring
on the router as well to aid with management. agent. Answers A, B, and C are incorrect because
Answers ii and iv are incorrect because you need they are all examples of server-monitoring agents.
the SNMP functionality available to your server, See Chapter 3, section “Documenting the
and it would not necessarily be so if installed on Network Server Configuration.”
the other choices. See Chapter 3, section “Install
23. A. Verification is a process that checks the content
Service Tools.”
of your backed-up media to the corresponding
18. A. Once the SNMP agent is installed, it must be files you attempted to back up. Answer B is not
configured with a SNMP community name. the best answer because the files you attempted to
Answers B, C, and D are incorrect because you back up might or might not be on a CD. Answer
do not have to configure any of these things with C is incorrect because it is a specific network
an SNMP agent. See Chapter 3, section “Install database that you might or might not have
Service Tools.” attempted to back up. Answer E is incorrect
because the tape vendor would have no idea what
19. A. A server baseline is a measurement of the server
you wanted to back up. See Chapter 4, section
performance. Answers B, C, and D are incorrect
“Perform Full Backup.”
because standards are diagnostics and partitions are
not measurements of performance. See Chapter 3, 24. B. Before you attempt any upgrade operation,
section “Performing a Server Baseline.” you should always fully back up the entire net-
work server. Answer A is incorrect because POST
20. A. A performance-monitoring tool can be used
tests are internal tests that run when you start
to ascertain the performance of various network
or reset your computer. Answer C is incorrect
server components under normal operating
because you do not need to fully back up a net-
conditions. Answer B is incorrect because an
work server upon each logoff. Answer D is incor-
antivirus program protects against viruses; it does
rect because the scheduling of full network server
not monitor performance. Answers C and D are
backups (daily, weekly, annually, and so on) is an
incorrect because “workload indexer” is a made-
administrative decision tailored to the backup
up term that in no way relates to monitoring
needs of the organization. See Chapter 4, section
system performance. See Chapter 3, section
“Perform Full Backup.”
“Performing a Server Baseline.”
25. D. To add another processor to a multiprocessor-
21. C. The Simple Network Management Protocol
capable network server, the new processor must
(SNMP) allows an SNMP management console
meet all four of the listed criteria. That is, the
to manage devices that have an SNMP agent
17 Serv+ Ch_PE 7/10/01 10:04 AM Page 539
processor must be the same model, have the same follow the rotational-speed standard when identi-
clock speed, have the same Level 2 cache size, fying disk drive speeds. Answer D is incorrect
and must match the stepping within 1 version. because drives rotate in one direction (unless
Answers A, B, and C are incorrect because they something is perilously wrong). See Chapter 4,
each just list two of these factors. See Chapter 4, section “Add Hard Drives.”
section “Add Processors.”
29. D. This mismatching of these metals causes cor-
26. C. If your original installation of Microsoft rosion at the contact points and results in a bad
Windows 2000 Server was on a network server connection over time. This bad connection causes
with a single processor, you must update the memory errors to occur. Answer A is incorrect
hardware abstraction layer (HAL) on the network because the mixing of these metals does not
server if you want the network server to recognize increase ESD. Answer B is incorrect because the
and use multiple processors. Answer A is incor- difference in cost is not prohibitive. Answer C is
rect because per-seat licenses relate to the number incorrect because the combination of these metals
of computers that are legally entitled to run an has nothing to do with application interoperability.
application. Answer B is incorrect because the See Chapter 4, section “Increase Memory.”
Add New Hardware icon is a Windows Control
30. B. The most reliable way to check the existing
Panel icon that enables you to add new devices,
memory configuration is to open the chassis of
to unplug/remove devices from your computer,
the network server. Answer A is incorrect because
and to troubleshoot hardware problems. Answer
there is no reason why the vendor should know
D is incorrect because there is no Multitasking
the memory specs of your network server. Answer
dialog box in Windows 2000 Server. See Chapter
C is incorrect because, as stated previously, the
4, section “Add Processors.”
vendor would have no reason to know the cur-
27. B. When adding additional disk drives, the best rent specs on your particular system. Answer D is
way to ensure interoperability with existing drives incorrect because you would not learn the memo-
is to install drives from the same manufacturer. ry configuration in this way (but you probably
Most manufacturers perform tests to make sure would learn a bit about recovery from system
that their own products operate together correctly. crashes). See Chapter 4, section “Increase
Answer A is incorrect because the DLLs do not Memory.”
reveal potential interoperability problems. Answer
31. D. The BIOS brand and version number usually
C is incorrect because the Help files will give you
display on the system monitor when the network
limited (if any) information about cross-platform
server is powered up. Answer A is incorrect
operability issues. Answer D is incorrect because
because this information is not usually written on
the IEEE 802 standard relates to local area net-
a sticker affixed to the back of the chassis. Answer
works, not to hard drive interoperability. See
B is incorrect because this information does not
Chapter 4, section “Add Hard Drives.”
display when you attempt to open a corrupted
28. A. Disk drive speeds are characterized by their file. Answer C is incorrect because the vendor’s
rotational speed. Answer B is incorrect because help desk would have no reason to know what
the RAM cache refers to the portion of RAM BIOS is on your system (unless you were able to
used as temporary workspace by various applica- identify it by version number, which is the infor-
tions (and is more related to processor speed). mation you’re actually looking for). See Chapter
Answer C is incorrect because drive manufacturers 4, section “Upgrade BIOS/Firmware.”
17 Serv+ Ch_PE 7/10/01 10:04 AM Page 540
32. C. The flash upgrade process is dangerous in that 36. C. External peripherals are defined as devices
the old BIOS code in the EEPROM is first erased external to the network server chassis, such as
before the new BIOS code is written to the EEP- printers, modems, monitors, keyboards, and mice.
ROM. Answer A is incorrect because the BIOS is Answer A is incorrect because the printer is an
erased. Answer B is incorrect because the erasure external peripheral. Answer B is incorrect because
occurs before, not after. Answer D is incorrect the monitor is an external peripheral. Answer D is
because the erasure doesn’t occur simultaneously incorrect because a modem is an external periph-
with the writing of the new BIOS. See Chapter eral. See Chapter 4, section “Upgrade Peripheral
4, section “Upgrade BIOS/firmware.” Devices, Internal and External.”
33. C. The term “hot expansion” refers to installing a 37. C. The system-monitoring agents usually use stan-
new adapter into a previously empty slot while the dard protocols such as HTTP, SNMP, and DMI
network server is running. Answer A is incorrect to report information to a management console.
because it is the definition of hot replacement. Answer A is incorrect because a network interface
Answer B is incorrect because it is the definition of card is an adapter that enables you to connect a
hot upgrade. Answer D is incorrect because closing network cable to a microcomputer. Answer B is
the slot release lever is one of the last steps in the incorrect because the IRQ channel is the line
process of adding an adapter by using PCI hot through which peripherals let a microprocessor
plug. See Chapter 4, section “Upgrade Adapters.” know they are ready to function. Answer D is
incorrect because even though an error log may be
34. B. A RAID controller uses onboard memory in
the information reported to the management con-
much the same way that a SCSI adapter uses
sole, it is not the means. See Chapter 4, section
memory, as a buffer. Answer A is incorrect because
“Upgrade System Monitoring Agents.”
the memory is not a network interface. Answer C
is incorrect because memory is not used as a 38. B. In some cases, the network-monitoring agents
firmware extension. Answer D is incorrect because are able to predict the impending failure of net-
RIMM is the RAMbus in-line memory module. work server components, such as the processor,
See Chapter 4, section “Upgrade Adapters.” the memory, or the hard disk drives. Answer A is
incorrect because the network monitoring agents
35. B. If the adapters are not Plug and Play, you may
cannot predict updates to system software.
have to configure the adapters with an interrupt
Answer C is incorrect because network-monitoring
(IRQ), a Direct Memory Access (DMA) channel,
agents monitor what is currently on the system,
and input/output (I/O) address. Answer A is
not those devices that will be added in the future.
incorrect because the flow control refers to the
Answer D is incorrect because although the
method of controlling the flow of data between
agents can report information to browsers, they
computers. Answer C is incorrect because a
have no capability to monitor so-called browser
USRT is a Universal Asynchronous Receiver.
wars. See Chapter 4, section “Upgrade System
Answer D is incorrect because a VLSI refers to a
Monitoring Agents.”
very large-scale integration, which relates to IC
devices with a high number of electronic compo- 39. D. All the listed choices are general categories
nents. See Chapter 4, section “Upgrade of service tools. See Chapter 4, section “Upgrade
Peripheral Devices, Internal and External.” Service Tools.”
17 Serv+ Ch_PE 7/10/01 10:04 AM Page 541
40. A. The utilities in the system/diagnostic partition 45. A. Performance Tool enables you to monitor
enable you to view and change the configuration system performance in Windows 2000. Answer
of the hardware components in the network server B is incorrect as this limits CPU time for web
without the network server operating system applications or sites in IIS. Answer C is incorrect
being loaded. Answers B, C, and D are incorrect as Event Viewer provides only an entry to the
because those utilities require that the operating log files. Answer D is incorrect as this provides
system be loaded. See Chapter 4, section system, user, and GDI resources information
“Upgrade Service Tools.” on local Windows 9x machines. See Chapter 5,
section “Working with Baselines.”
41. B. If the UPS has to be turned off, the network
server must be shut down for the UPS battery 46. D. A baseline is a snapshot in time of how a serv-
replacement procedure. Answer A is incorrect er is running and enables you to compare system
because the modem does not affect the replace- performance over time, and note any deviations
ment of a UPS. Answer C is incorrect because in performance after changes are made. Answer A
you do not need to reboot the network server is incorrect as this is done by Net Watcher or a
before making the replacement. Answer D is similar utility. Answer B is incorrect as you can
incorrect because you do not have to call the help monitor memory and processor objects, but can-
desk; you can usually get the information you not troubleshoot them with baseline data. Answer
need from the Installation Manual or online C is incorrect as this is done within Performance
from the vendor’s web site. See Chapter 4, section Tool to gather baseline data. See Chapter 5, sec-
“Upgrade Uninterruptible Power Supply.” tion “Working with Baselines.”
42. D. Among other questions, you must ask all those 47. C. The SNMP service runs at UDP port 161 by
that are listed in this question. See Chapter 4, sec- default. Answer A is incorrect as this is the port
tion “Upgrade Uninterruptible Power Supply.” used for the WWW service. Answer B is incorrect
as this is the port used by the email service
43. C. A “regular” backup schedule is determined by
POP3. Answer D is incorrect as it is the port
the value of your data and the type of business
used by FTP. See Chapter 5, section “Working
you operate. One company’s “regular” backup
with SNMP.”
schedule could be every day, whereas another
company’s “regular” backup schedule could by 48. C. You can request the next object’s value with
every month. Answers A, B, and D are incorrect get-next, such as query a client’s ARP cache and
because these intervals may not be right for your then ask for each subsequent value. Answer A is
company and the value of your data. See incorrect as this describes the get command.
Chapter 5, section “Understanding Backups.” Answer B is incorrect as this describes the set
command. Answer D is incorrect as this is the v
44. B. Optional is not a normal backup category.
option when using the tar utility in Linux. See
Answers A, B, C are incorrect because they are
Chapter 5, section “Working with SNMP.”
normal backup categories marketed in software
utilities and tools. See Chapter 5, section 49. C. A plenum is the space between the ceiling and
“Understanding Backups.” the floor of a building’s next level. Answer A is
incorrect as this is usually polyvinyl chloride.
17 Serv+ Ch_PE 7/10/01 10:04 AM Page 542
Answer B is incorrect as this area does not have a Answer C is incorrect as this done by Microsoft
title. Answer D is incorrect as EMI is electromag- Backup. Answer D is incorrect as this is done by
netic interference. See Chapter 5, section “The Backup or other backup utility. See Chapter 5,
Physical Environment.” section “Establishing Remote Notification.”
50. B. PVC is prohibited as it gives off poisonous 55. C. In a large environment, using a lock-and-key
fumes when it burns. Answer A is incorrect as method to secure a server can be ineffective if
plenum-grade cable is allowed in the plenum. locks need rekeying and keys need to be reissued
Answer C is incorrect as this is the type of cable a number of times. Answer A is incorrect as a
that is allowed by fire code. Answer D is incorrect combination lock can be reset at any time.
as this is not a plenum-grade cable. See Chapter Answers B and D are incorrect as you can easily
5, section “The Physical Environment.” authorize and remove access from a central
authentication server. See Chapter 6, section
51. B. Device Manager provides detailed information
“Physical Security Issues.”
about COM ports and other device information.
Answer A is incorrect because it provides general 56. A. Most high-end servers include locking cases
system information, including the processor and and the ability to lock the drives into the server
RAM. Answer C is incorrect because SNMP chassis. Answers B, C, and D are incorrect as
reports network usage statistics. Answer D is these components do not lock. See Chapter 6,
incorrect as it is used to gather system performance. section “Physical Security Issues.”
See Chapter 5, section “Verifying Hardware.”
57. A. A UPS or surge suppressor should be used
52. C. The proc file system structure can provide you on all servers to alleviate power-line issues.
with information about your system. Answer A Answer iii is incorrect as antistatic devices are
is incorrect as this is the Linux utility for copy used to eliminate ESD. Answer iv is incorrect
in/out for backups. Answers B and D are incorrect as it is the voltage at which a surge suppressor
as these are backup commands for performing reacts to a spike or surge. See Chapter 6, sec-
backups in Linux. See Chapter 5, section tion “Environmental Issues.”
“Verifying Hardware.”
58. B. Fire suppression, ESD, and air quality should
53. B. is the Simple Network Management Protocol also be examined when establishing a safe server
and can be used on TCP/IP networks. Answer A environment. Answer ii is incorrect as sound
is incorrect as this is the Windows NT Hardware should not interfere with server safety. See
Query tool. Answer C is incorrect as this is the Chapter 6, section “Environmental Issues.”
File Transfer Protocol and cannot notify you of
59. C. A faulty terminator will not stop messages as
situations that arise. Answer D is incorrect as this
they reach the end of the run, allowing them to
is polyvinyl chloride. See Chapter 5, section
loop back and increase traffic along the wire (also
“Establishing Remote Notification.”
increasing the number of collisions). Tape media
54. A. The Messenger service can send you alerts can hold data in compressed and uncompressed
based on thresholds you’ve established on the net- format. Answer A is incorrect as a 10BASE-2 seg-
work. Answer B is incorrect as this is done by ment can hold 30 nodes. Answer B is incorrect as
Device Manager, the BIOS, and other utilities. hubs are not used with 10BASE-2. Answer D is
17 Serv+ Ch_PE 7/10/01 10:04 AM Page 543
incorrect as transceivers are used with 10BASE-5, 64. C. Print jobs are spooled to the hard drive. If the
but not with 10BASE-2 networks. See Chapter 7, hard drives are small, users will be unable to add
section “Perform Problem Determination.” jobs when the space is gone. Answers A, B, and
D are incorrect as they do little with assisting
60. C. With 10BASE-2, a disconnected cable can
users with print job submission. See Chapter 7,
prevent all network traffic from traversing the
section “Identify Bottlenecks.”
wire. If the client is running a Windows-based
operating system and a DHCP server is used to 65. A. The workgroup name in this scenario would
issue IP addresses, Automatic IP Addresses be using NetBEUI broadcasts to build the list of
(AIPA) will use addresses in the 169.254.x.x available resources. Without the correct work-
range if the DHCP server cannot be reached. The group name, the resources may not be viewable.
unhooked cable can also prevent printing to, or Answer B is incorrect as frame types are used in
accessing other network resources. Answer iv is NetWare networks, but not this scenario. Answer
incorrect because the inability to communicate C is a possibility, but not the most likely cause.
on the network should not affect other operations Answer D is incorrect as the authentication pro-
at all. See Chapter 7, section “Perform tocol is not needed in the workgroup setting, and
Problem Determination.” improperly configuring would not prevent seeing
the resources. See Chapter 7, section “Identify
61. B. The config command is used on NetWare to
and Correct Misconfigurations and/or Upgrades.”
see the IP stack configuration parameters. Answer
A is incorrect as this command is used on UNIX 66. D. The number of licenses for the application
and Linux systems. Answer C is incorrect as this being exceeded will not prevent you from seeing
command is used on Windows NT and 2000 the server. Answer A will prevent you from com-
systems. Answer D is incorrect as this command municating after your lease expires. Answer B will
is used on Windows 95 and 98 systems. See prevent you from seeing the server. Answer C
Chapter 7, section “Use Diagnostic Hardware will prevent you from communicating outside
and Software Tools and Utilities.” the subnet. See Chapter 7, section “Identify and
Correct Misconfigurations and/or Upgrades.”
62. D. Six, as this will shut the system down and
reboot it. Answer A is incorrect as it will shut the 67. B. When an application is installed, it includes a
system down but not reboot it. Answer B will number of ancillary files needed to make it work.
take the server to administrative mode, but not If those files are also used by other applications, it
reboot. Answer C is the normal operating mode. is possible that newer versions can adversely affect
See Chapter 7, section “Use Diagnostic Hardware operations of the other applications. While answers
and Software Tools and Utilities.” A, C, and D are all possible, they are not the
most likely cause and would also affect issues
63. B. Running applications across a network
other than those presented. See Chapter 7, sec-
requires a great deal of memory. Answers A, C,
tion “Determine If Problem Is Hardware,
and D are incorrect as they would not be as
Software, or Virus Related.”
important in this scenario as RAM. See Chapter
7, section “Identify Bottlenecks.”
17 Serv+ Ch_PE 7/10/01 10:04 AM Page 544
68. B. A virus will cause the symptoms described. 73. A. Vendor web sites contain the most up-to-date
Answer A will prevent the system from reaching hardware compatibility lists for network operat-
this point. Answer C is not a likely scenario and ing systems. Answers B, C, and D are incorrect as
would not affect printing. Answer D would the information may become obsolete in a short
not affect printing. See Chapter 7, section time because of operating system updates and
“Determine If Problem Is Hardware, Software, changes. See Chapter 2, section “Verifying
or Virus Related.” Hardware Compatibility with the Operating
System.”
69. C. In a RAID 1 array, each drive is completely
mirrored to another drive. Answer A is incorrect 74. C. A volt/ohm meter or a digital multimeter are
as this applies to RAID 5. Answer B is incorrect as used to check the voltage of the power source.
this applies to RAID 0. Answer D is incorrect as it Answer i. is incorrect as it is used as a back up
applies to RAID 0+1. See Chapter 8, section power source in a UPS. Answer ii is incorrect as
“Creating a Disaster-Recovery Plan.” it allows the keyboard, mouse, and video display
to be switched among network servers. See
70. D. A is incorrect as RAID 0 requires N drives
Chapter 2, section “Verifying the Power Source.”
to implement, where N is the number of drives.
Answer B is incorrect as RAID Level 0+1 requires 75. B. The time range for an adequate size UPS to
a minimum of 4 drives to implement. Answer C keep the network server running in order to per-
is incorrect as you will need a minimum of two form a normal shutdown is 5 to 10 minutes.
drives to implement RAID 1. See Chapter 8, Answer A is considered undersized and will only
section “Creating a Disaster-Recovery Plan.” keep the server running a very short period of
time. Answers C and D are considered oversized
71. B. A hot site is a commercial facility that is avail-
time ranges. See Chapter 2, section “Verifying the
able for system backups. Answer A is incorrect
Uninterruptible Power Supply Size.”
because a cold site is a facility that has been pre-
pared to receive computer equipment in the event 76. B. A server rack saves space, provides a secure
that the primary facility is destroyed or becomes environment for equipment, and is designed to
unusable. Answers C and D are incorrect because aid in cooling. Answers A, C, and D are incorrect
they are just made-up terms. See Chapter 8, section as all equipment must be designed to fit into a
“Implementing Your Disaster-Recovery Plan.” server rack. See Chapter 2, section “Server Rack.”
72. B. For memory, you want to verify that the type of 77. A. The RAID controller must be configured
memory is correct (for instance, error-correcting before the network operating system can be
memory, non-error correcting, parity). You also installed. Answer B is incorrect as the configura-
should check the speed of the memory (50 tion must occur prior to installing the network
nanosecond, 60 nanosecond, and so on), and operating system. Answer C is incorrect as when
finally check the form factor. Answer iv is incor- the system is placed in a server rack should not
rect because Ethernet is a cabling standard unre- matter. Answer D is incorrect as a DVD-ROM
lated to memory. See Chapter 8, section drive can be added at any time. See Chapter 2,
“Implementing Your Disaster-Recovery Plan.” section “RAID Controller(s).”
17 Serv+ Ch_PE 7/10/01 10:04 AM Page 545
III
P A R T
APPENDIXES
A Glossary
A P P E N D I X A
Glossary
100BASETX Ethernet standard that uses unshielded adaptive fault tolerance An adapter teaming tech-
twisted-pair (UTP) cable with a maximum cable length nique in which multiple NICs are used to act as backup
of 100 meters from the NIC to the hub. Operates at for one another in case of a NIC failure.
100 megabits per second.
Advanced Intelligent Tape A tape technology that
10BASE2 Ethernet standard that uses thin coaxial uses 8mm videotapes and an onboard memory.
cable (RG-58) with a maximum length of 185 meters.
American National Standards Institute (ANSI)
Operates at 10 megabits per second.
Organization that defines standards for the United
10BASE5 Ethernet standard that uses thick coaxial States, many of which deal with network communica-
cable (RG-8) with a maximum length of 500 meters. tions and computers.
Operates at 10 megabits per second.
American Standard Code for Information
10BASEF Ethernet standard that uses fiber-optic Interchange (ASCII) 7- or 8-bit code that defines
cable. Operates at 10 megabits per second. the representation of the alphabet, punctuation,
numerals, and other symbols used in computers.
10BASET Ethernet standard that uses unshielded
twisted-pair (UTP) cable with a maximum cable length ANSI American National Standards Institute.
of 100 meters from the NIC to the hub. Operates at
antivirus software Software that scans the network
10 megabits per second.
server for files containing viruses. It will then either
802.3 IEEE standard that defines Ethernet networks. isolate or remove the viruses.
802.5 IEEE standard that defines Token Ring APIPA (Automatic Private Internet Protocol
networks. Addressing) A method of assigning an IP address
automatically when a DHCP server is not available on
802.11 IEEE standard that defines Wireless networks.
a network. The IP address assigned will be on the
169.254.0.0 network.
archive bit A flag indicating whether a file has been
A created or modified since the last full or incremental
backup was performed.
adapter load balancing An adapter teaming tech-
nique where multiple NICs are used to share the load ATA (Advanced Technology Attachment) The
to and from the network server. ANSI name for the IDE specification.
adapter teaming The technique of using multiple ATA-2 (Advanced Technology Attachment-2) The
NICs together, either for load balancing or fault ANSI name for the EIDE specification.
tolerance.
19 Serv+ Ch_AppA 7/10/01 10:05 AM Page 550
B C
backup software Software that will copy files from cache A high-speed memory that stores frequently
network server disk drives to a backup device, such as used data and/or instructions that operate between the
a magnetic tape drive. processor and the system memory. Also a temporary
storage area between the disk drives and the processor.
baseline A snapshot of system performance against
which future performance can be measured. clamping speed The speed at which a surge suppres-
sor reacts to a power spike or surge.
Berkeley Internet Name Domain (BIND) A DNS
server often used on the UNIX operating system. clamping voltage The voltage at which the surge
suppressor reacts to a spike or surge.
biometrics Authentication techniques that use
measurable physical characteristics for authentication. cluster A group of disk sectors. The smallest amount
Fingerprints are an example of a biometrics-based of space that be allocated on a disk drive.
authentication mechanism.
clustering Connecting two or more computers
BIOS (Basic Input/Output System) A program that together so that they act as a single computer.
resides in firmware that controls all low-level functions
cross-over cable An Ethernet unshielded twisted-pair
(for instance, starting up and access to devices) on
cable that has wires 1 and 2 crossed with wires 3 and 6.
either the system board or on the individual compo-
This nonstandard cable can be used to connect two
nents in the network server, such as a SCSI controller
Ethernet hubs together.
or a RAID controller.
blackout A blackout (power outage) occurs when
there is no power being delivered to the server. These
can vary from an instant to an extended period of time.
D
British Thermal Unit (BTU) A measure of the daemon The named used for a process or service on
quantity of heat, defined since 1956 as approximately the UNIX operating system.
equal to 1,055 joules, or 252 gram calories.
departmental server A network server designed to
brownout An undervoltage condition that lasts for support a relatively small number of clients on a net-
extended periods of time. work.
buffer A memory storage area between devices of dif- Desktop Management Interface (DMI) A standard
ferent speeds, such as the computer’s memory and its by the Desktop Management Task Force (DTMF) to
disk drives. allow for the monitoring of desktop computers.
buffered memory EDO memory with special diagnostic partition A special disk partition created
circuitry to support a large amount of memory in a by some network server hardware vendors that contains
network server. diagnostics for troubleshooting the network server when
the network server operating system isn’t available.
bus A parallel set of conductors that carry data from
one location to another in a computer. differential backup A backup of all files on the
system that have been created or changed since the
last full backup.
19 Serv+ Ch_AppA 7/10/01 10:05 AM Page 551
differential SCSI Also known as high-voltage differ- disk mirroring A fault-tolerance technique that stores
ential (HVD). A SCSI signaling system that allows a the same data on two disk drives that are attached to the
total SCSI cable length of 25 meters. HVD is not same disk controller. One of two RAID 1 techniques.
compatible with single-ended (SE) devices.
disk striping with parity A fault-tolerance technique
Digital Audio Tape (DAT) A tape backup system that that writes data to multiple disks with parity information.
uses digital audio tape as a backup medium. Uses Digital
disk striping without parity A technique that is not
Data Storage (DDS) as the data recording system.
fault tolerant for writing data to multiple disks at the
Digital Data Storage (DDS) The recording system same time, improving disk input/output performance.
used by Digital Audio Tape.
Distributed Lock Manager (DLM) Application
Digital Linear Tape (DLT) A tape backup system used in clustering to handle resource contention in a
that records data along linear tracks on a tape. shared resource clustering model.
DIMM (Dual In-line Memory Module) A small cir- DMI (Desktop Management Interface) A standard
cuit board that can hold a group of memory chips, typi- for centrally managing and keeping track of hardware
cally used as computer memory. DIMMs have a 168-pin and software components in a group of microcomputers.
connector. DIMMs have a 64-bit memory path, making
Domain Name Service (DNS) A network service
them well suited for use with Pentium processors.
that resolves TCP/IP names to TCP/IP addresses.
Direct Memory Access (DMA) A methodology that
dual in-line memory module (DIMM) A memory
allows data to be transferred from disk drives or NICs
module with 168 pins used in network servers.
directly to system memory without the processor
being involved. Dynamic Host Configuration Protocol (DHCP) A
protocol that provides automatic configuration of
directory services server A network server that stores
TCP/IP for client computers.
and manages the network directory information.
dynamic random access memory (DRAM)
disaster-recovery plan Process developed to ensure
Memory that must be refreshed frequently for it to
computing operations can be resumed in the event of
“remember” the data stored in it.
an outage.
disk array A group of disk drives that perform togeth-
er as a single entity.
E
disk duplexing A method for adding fault tolerance
to drive subsystems. Duplexing duplicates information EEPROM (electrically erasable programmable read-
from one set of drives to another (similar to mirroring), only memory) Memory often used to contain a
so that the server always has an exact copy of the data “flash BIOS.” Erasable by applying high voltage levels.
in two different places. Duplexing, however, increases
EIDE (Enhanced Integrated Drive Electronics, or
the redundancy by implementing two drive controllers,
Enhanced IDE) An enhanced version of the IDE that
so that the system can survive not only the loss of a
supports storage devices, such as CD-ROMs, tape drives,
drive, but also the loss of a controller. Disk duplexing
and hard disk drives at maximum transfer rates of up to
is one of two RAID 1 techniques.
16.6MBps. Adopted by ANSI as ATA-2 in 1994.
19 Serv+ Ch_AppA 7/10/01 10:05 AM Page 552
electrostatic discharge (ESD) ESD is the rapid FAT32 File allocation table file system that uses 32
discharge of static electricity from one conductor to bits to address clusters on a disk drive. FAT32 can
another. An electrostatic discharge (static shock) can address disk drive partitions up to 8GB using the
damage computer circuits. smallest cluster size of 4KB. Disk partitions up to 2
terabytes can be addressed using larger cluster sizes.
enterprise server A network server capable of sup-
porting hundreds of users. FAX server A network server that enables clients to
send and receive facsimiles over the network.
EPROM (erasable programmable read-only memo-
ry) Memory that can be programmed once, but it Fibre Channel A communications system that can
can be erased by exposure to ultraviolet light and be used to build storage area networks (SANs). A form
then reprogrammed. of serial SCSI.
Error Correcting Code (ECC) A technique for Fibre Channel Arbitrated Loop (FC-AL) The most
detecting and correcting a single-bit error in memory popular form of Fibre Channel that connects devices
and detecting a double-bit error. together in a loop.
Ethernet A network communications system created File Transfer Protocol (FTP) A TCP/IP protocol
by Robert Metcalfe and standardized by IEEE as 802.3. that allows files to be transferred between computers.
Operates at 10 megabits per second.
FireWire (IEEE 1394) A form of serial SCSI that
Event Viewer The Windows NT Server and supports Plug and Play and hot plugging for up to 63
Windows 2000 Server utility that allows viewing of devices at 400Mbps.
the System, Security, and Applications log files.
firmware Halfway between software and hardware,
Extended Binary Coded Decimal Information Code firmware is a program (software) that is “burned into”
(EBCDIC) An 8-bit code used to represent the letters a programmable read-only memory module. This
of the alphabet, numerals, and punctuation marks. firmware usually controls a particular function on an
Used primarily on IBM mainframes. adapter board in a network server. If an EEPROM is
used for the firmware, it can be reprogrammed.
flash BIOS upgrade A process of reprogramming
F a BIOS that is stored in “flash” memory (usually
an EEPROM).
Fast Ethernet An enhancement to the original
flashable BIOS A Basic Input/Output System
Ethernet that operates at 100 megabits per second.
(BIOS) that can be changed by running a program that
FAT (file allocation table) The file system intro- erases the old BIOS and writes a new BIOS.
duced with early versions of DOS. Nearly all micro-
Frequently Asked Questions (FAQs) A complica-
computer operating systems since DOS can read and
tion of answers to questions that are often asked of
write files written on FAT file systems.
either technical support providers or in newsgroups
FAT16 File allocation table file system that uses 16 dedicated to a particular subject.
bits to address clusters on a disk drive. Fat16 can
FRU (field replaceable unit) A component of the
address disk drive partitions up to 2GB.
network server that can be replaced at the customer’s
site (in the field).
19 Serv+ Ch_AppA 7/10/01 10:05 AM Page 553
full backup A backup where all of the files on the HTTP (Hypertext Transport Protocol) The net-
network server are written to tape. work protocol used by web browsers and web servers to
communicate with each other.
Hypertext Transfer Protocol (HTTP) The TCP/IP
G protocol that enables web browsers to talk to
web servers.
gigabit (Gb) One billion bits (109) or 1,073,741,824
bits (230) bits depending on context.
Gigabit Ethernet An enhancement to the Ethernet I
standard that operates at 1000Mbps (1 gigabit/second).
IBM OS/2 An IBM network server operating system.
gigabyte (GB) One billion bytes (109) or
1,073,741,824 bytes (230) bytes depending on context. IDE (Intelligent Drive Electronics) IDE is an inter-
face between the computer bus and the disk controller
that is integrated on the disk drive. The IDE specifica-
tion was adopted as the ATA standard by ANSI in
H November 1990.
hardware RAID A RAID system implemented by IEEE 1394 (FireWire) A form of serial SCSI that
using a RAID controller to manage the disk arrays. supports Plug and Play and hot plugging for up to 63
at 400Mbps.
high-voltage differential (HVD) SCSI A SCSI sig-
naling system that allows cable lengths of up to ifconfig The UNIX command that will display the
25 meters. TCP/IP configuration for a network interface card.
hostname A TCP/IP command that will display the incremental backup A backup scheme in which only
name of the network server. the files that were created or changed since the last full or
incremental backup are written to the backup medium.
hot spare A disk drive that is powered up and is part
of a RAID system that is not utilized until one of the InfiniBand Architecture (IBA) A proposed high-
disk drives in a RAID array (RAID 1 or RAID 5) fails. speed input/output system for network servers.
Upon failure of one of the disk drives, the hot spare is
Institute of Electrical and Electronic Engineers
automatically utilized to rebuild the RAID array. Also
(IEEE) An association that sets standards, many of
used to refer to any spare component or server kept
which deal with computers and networks.
powered up and available as a replacement part in
the event that a part fails. Can include entire servers Intelligent Input/Output (I2O) An improved
if applicable. input/output system for network servers that removes
the burden of input/output from the processor.
hot swap A drive technology that allows drives to be
swapped in a system without requiring the suspension Internet Message Access Protocol (IMAP) An email
of activity. protocol that transfers email messages between an
SNMP server and an email client.
HPFS (High Performance File System) File system
used by IBM’s OS/2 operating system.
19 Serv+ Ch_AppA 7/10/01 10:05 AM Page 554
iostat (Input/Output Statistics) A UNIX command Linux A free, open-source, network operating system
that reports disk input/output activity. This information that is a clone of UNIX.
is useful in finding disk subsystem bottlenecks.
low-voltage differential (LVD) A SCSI signaling sys-
ipconfig The Windows NT/Windows 2000 com- tem that allows a maximum cable length of 12 meters
mand that will display the TCP/IP configuration of the and higher speeds than single-ended signaling systems.
network server.
LUN (Logical Unit Number) A mechanism that
IPX/SPX protocol A network protocol popularized allows several SCSI devices to use the same SCSI ID.
by Novell in its NetWare network operating system.
M
K
magic packet The special TCP/IP packet that is sent
keyboard/video/mouse (KVM) switch A device that to a NIC with Wake-On-LAN technology that will
allows a single keyboard, video display, and mouse to signal the NIC to power up the computer.
be shared among several network servers.
man A UNIX command that can be used to get
kilobit (Kb) One thousand (103) bits or 1,024 bits information about UNIX. This online documentation
(210) depending on the context. commands (known as man pages) is useful to ascertain
the flags for commands that vary among the versions
kilobyte (KB) One thousand (103) bytes or 1,024
of UNIX.
bytes (210) depending on the context.
megabit (Mb) One million bits (106) or 1,048,576
KVM switch (keyboard, video, mouse switch) A
bits (220) depending on the context.
KVM switch allows the use of a single keyboard,
video display, and mouse to control several computers megabyte (MB) One million bytes (106) or
(servers) by means of switching the three devices 1,048,576 bytes (220) depending on the context.
among the servers.
megahertz (MHz) One million hertz (cycles
per second).
MIB (Management Information Base) A database
L of the items that can be managed by an SNMP agent
running on a managed network device.
L2 cache Level 2 microprocessor cache. Level 2 cache
is placed between Level 1 cache (on the processor chip Microsoft Windows 2000 Server A Microsoft net-
itself ) and system memory. Today, Level 2 cache is usu- work operating system built on Windows NT technolo-
ally located on the processor chip along with the Level 1 gy that includes a directory service.
cache. However, the Level 2 cache is usually much larger
Microsoft Windows NT Server A Microsoft net-
than Level 1 cache.
work operating system.
Linear Tape–Open (LTO) A tape backup technology
that promises high storage capacity or high input/
output speed.
19 Serv+ Ch_AppA 7/10/01 10:05 AM Page 555
N P
nbtstat The Windows NT/Windows 2000 com- paging file The disk space used to augment physical
mand that will display information about NetBIOS system memory in an operating system that supports
over TCP/IP sessions on the network server. virtual memory. Another name for a swap file.
NetBEUI protocol An early network protocol used partition A logical portion of a disk drive that will be
by IBM and Microsoft. identified by a drive letter or volume name.
NetBIOS A network naming system that is part of pathping A Windows 2000 network diagnostic tool
the NetBEUI protocol. that combines the Ping and Tracert utilities.
netstat The Windows NT/Windows 2000 and PCI bus Peripheral Component Interconnect local
UNIX utility that will display statistics about TCP/IP bus introduced by Intel in 1992. A PCI bus operates at
connections on the network server. 33MHz and can transfer 32 bytes at a time. This gives
adapters plugged into the PCI bus a throughput of
network-attached storage (NAS) A network storage
132MB per second.
system that is available via standard network protocols
and available on all computers on the network. PCI hot plug A technology that allows for the instal-
lation, replacement, and upgrades to PCI adapters
network interface card (NIC) Computer card used
without shutting down the network server.
to connect a computer to a network.
peer-to-peer network A network system where each
non-ECC memory Memory that does not have
computer can be both a client and a server at the
Error Correcting Code technology.
same time.
Novell NetWare A network operating system created
ping The TCP/IP utility used to test TCP/IP con-
by Novell.
nectivity between computers on a TCP/IP network.
nslookup The TCP/IP utility used to test name reso-
Plug and Play A technology that allows for the auto-
lution on a DNS server.
matic detection and configuration of components as
NSS (Novell Storage Services) The newest file sys- they are attached to or installed in the network server
tem used on Novell’s latest release of NetWare, version or client computer.
5.
port aggregation A technique that uses several net-
NTFS File system used by Microsoft Windows NT work ports together to act as a single high-speed port.
Server and Microsoft Windows 2000 Server. NTFS
POST Diagnostics Board A special board that will
supports file compression, file and folder permissions,
display error message numbers when a network server
and long filenames.
fails to boot.
NWFS (NetWare File System) The file system used
Post Office Protocol (POP) An email protocol that
on all versions of Novell’s NetWare up to and including
transfers email messages between an SNMP server and
version 5.
an email client.
NWLink protocol Microsoft’s network protocol that
is compatible with Novell’s IPX/SPX protocol.
19 Serv+ Ch_AppA 7/10/01 10:05 AM Page 556
proactive maintenance Maintenance that is done to RAID 0 Referred to as drive striping, it is imple-
prevent problems from occurring or make it easier to mented by writing data sequentially across a series of
respond to them when they do occur. drives. This configuration provides enhanced disk
performance, but offers no redundancy.
programmed input/output (PIO) The process used
by IDE disk systems that requires the computer proces- RAID 0/1 A stripe set using RAID Level 0 that is
sor to perform the input/output operation. mirrored (RAID Level 1). Requires at least four disk
drives to implement.
ps A UNIX/Linux command that will display the
processes that are currently running on the UNIX/ RAID 0+1 Another name for RAID 0/1.
Linux network server.
RAID 1 Referred to as drive mirroring, it is imple-
mented by writing identical data to two drives concur-
rently. In the event that one drive fails, the other drive
Q containing the “mirror image” data, can pick right up
and continue processing.
Quarter Inch Cartridge (QIC) An early tape backup
RAID 2 A technique requiring at least three disk
technology that was slow by current standards and of
drives and a hamming code to provide fault tolerance.
low storage capacity.
Never implemented commercially.
RAID 3 A technique requiring at least three disk
drives and using a bit-based parity to provide fault tol-
R erance. Parity information is maintained on a separate
disk drive from the data.
rack A rack is a large cabinet used to hold electronic
equipment, such as servers, networking equipment, RAID 4 A technique requiring at least three disk drives
routers, hubs, and so on. The rack is usually 6 to 7 feet and using a block-based parity to provide fault tolerance.
in height, and an industry-standard exists for the width Parity information is maintained on a separate disk
(19 inches), allowing equipment from different manu- drive from the data.
facturers to be mounted in the rack. Rack height is
RAID 5 Referred to as drive striping with parity, it
normally specified in rack units (U) (1.75 inches). A
uses the same basic technique as drive striping, but
42U rack is 73.5˝ (6´ 1.5˝) high.
instead of just writing data to each of the disks, a parity
RAID (redundant array of inexpensive disks) The bit is written with each data write. This parity data can
strategy of adding redundancy or performance to drive then be used to calculate the value of data in the event
subsystems by striping the data across multiple drives or a drive fails. Requires at least three drives.
by writing the same data to two different disk drives. By
RAID 10 Another name for RAID 0/1.
adding parity, customers can help protect their data,
allowing the system to still retrieve information despite a RAID controller BIOS The fundamental program
drive failure. Used to help ensure high server availability. that controls low-level operations on a RAID controller.
random access memory (RAM) The type of memo-
ry used by network servers for main memory.
19 Serv+ Ch_AppA 7/10/01 10:05 AM Page 557
redundancy The technique for ensuring availability SCSI Accessed Fault-Tolerant Enclosures (SAF-TE)
by having duplicate hardware available. A specification for monitoring SCSI devices in an
external enclosure.
redundant array of inexpensive disks (RAID) See
RAID. SCSI controller BIOS The fundamental program
that controls low-level operations on a SCSI controller.
redundant array of inexpensive tapes (RAIT)
RAID technology extended to tape drives. SCSI Enclosure Services (SES) A set of commands
that are defined in the SCSI-3 standard that allow
registered memory SDRAM memory with special
management of SCSI or FC-AL enclosures and the
circuitry to support a large amount of memory in a
devices that they contain.
network server.
SCSI-1 New name for the old SCSI, needed since
remote access server (RAS) A network server that
SCSI-2 and SCSI-3 have been introduced. SCSI-1 is
provides dial-in access to a network.
an 8-bit bus with a transfer speed of 5MBps. Can have
RIMM (Rambus memory module) A memory mod- up to 7 devices.
ule with either a 168- or 184-pin connector. RIMMs
SCSI-2 Defined in 1994, SCSI-2 has 8-bit and 16-
use RDRAM (direct Rambus memory) rather than the
bit buses and can transfer data at 10MBps or 20MBps.
SDRAM used on DIMMs.
Can have up to 7 devices on an 8-bit SCSI channel, 14
risk analysis The process by which the threats to a devices on a 16-bit (Wide) SCSI channel.
resource are analyzed and categorized based on severity
SCSI-3 The third and lastest SCSI standard, which is
and likelihood.
actually a set of standards for parallel SCSI and serial
RJ-45 A modular plug used to terminate 4-pair UTP SCSI. Speeds as high as 320MBps are defined.
cable for networking applications. The RJ-45 connector
sector A 512 byte portion of a disk drive.
plugs in a network interface card in a computer on one
end and a hub on the other end of a network cable. Self-Monitoring, Analysis and Reporting Technology
(S.M.A.R.T.) A technology for monitoring disk drives
that can predict impending failure of the disk drive.
S Serial Storage Architecture (SSA) A fast external
system for creating storage area networks that was
sag An undervoltage condition that is corrected in a created by IBM.
few milliseconds.
server appliance A server designed to deliver one
sar (System Activity Reporter) A UNIX command resource or one service to the network.
that generates a report of system activity that can be
server-centric network A network that has one or
used to analyze system performance.
more servers to provide resources and services. All other
scalability Technology that ensures the available computers on the network are clients.
resources of a system can handle additional load
service tools Software supplied by the network server
through either excess capacity or upgrades.
hardware vendor and the network operating system
SCSI (Small Computer Systems Interface) (pro- vendor to facilitate the configuration, upgrading, main-
nounced “skuzzy”) SCSI is a parallel interface tenance, and troubleshooting of a network server.
defined by ANSI.
19 Serv+ Ch_AppA 7/10/01 10:05 AM Page 558
shielded twisted-pair (STP) A type of wiring used static random access memory (SRAM) Memory
to create early Token Ring networks. that does not require frequent refreshing in order to
maintain data stored in it.
shutdown The process of gracefully powering down a
network server. Also the name of the command to initiate stepping Intel’s term for versions of their processors.
a shutdown on some network server operating systems. Each time there is a change to the processor, Intel
assigns a different stepping to it.
SIMM (single in-line memory module) A memory
module with a 72-pin connector and a 32-bit data path. storage area network (SAN) A network storage sys-
tem, implemented by FC-AL or SSA, that is available
Simple Mail Transfer Protocol (SMTP) A TCP/IP
to network servers on a separate network from the net-
protocol for delivering email messages.
work used by client computers.
Simple Network Management Protocol (SNMP) A
STP (shielded twisted-pair) Cable used in Token
TCP/IP protocol designed to allow management of
Ring networks. Uses a DB-9 connector on one end and
network devices.
a Universal Data Connector on the other end.
single-ended SCSI The original SCSI signaling system.
straight-through cable An unshielded twisted-pair
However, it has limited SCSI cable length (6 meters) and
cable where wires 1–8 connect to wires 1–8 on each
is usable only at relatively low speed.
end. This is the standard cable used to connect and
single in-line memory module (SIMM) A memory NIC to a hub.
module with 72 pins and a 32-bit data path.
striped set with parity Also known as a RAID 5 disk
single large expensive disk (SLED) The opposite array. An array of disk drives with built-in redundancy
of RAID. that can survive the failure of one of the disk drives in
Small Computer Systems Interface (SCSI) A bus the array.
controller system that allows many devices to be striped set without parity Also known as a RAID 0
attached to a single channel. disk array. This array provides no redundancy, but it
SNMP (Simple Network Management Protocol) A does improve disk input and output performance.
TCP/IP protocol used to manage network devices. surge If an overvoltage condition can be measured in
SNMP community A primitive security system used milliseconds, it is considered a surge.
by the SNMP protocol. swap file The disk space used to augment physical
software RAID RAID implemented by software in system memory in an operating system that supports
a network operating system that requires the use of virtual memory. Another name for a paging file.
the network server’s processor to implement the system BIOS A program that resides in firmware
RAID technology. that controls all low-level input and output among net-
spike An overvoltage condition can be measured in work server components.
nanoseconds, this condition is considered a spike. system monitoring agents Network server vendor-
standard rack unit In a standard 19-inch electronic specific software that monitor the components in the
equipment rack, a standard rack unit is 1.75 inches network server and report the status to a vendor-
in height. supplied monitoring console.
19 Serv+ Ch_AppA 7/10/01 10:05 AM Page 559
A P P E N D I X B
Overview of the
Certification Process
In addition, the role of server hardware specialist has á Demonstrates high levels of leadership through
evolved into a clearly defined and crucial one in IT mentoring and training others in server concepts
departments and organizations. CompTIA describes and operations.
the role as follows:
You may be asking why this exam is for you, and why
á Has in-depth knowledge of servers, including
now. Aside from the fact that the certification brings
working knowledge of troubleshooting, physical
certain obvious professional benefits to you, the Server+
security, and disaster recovery. Can recover from a
program gives you access to the CompTIA organization
server failure.
and to the benefits this access affords. CompTIA’s Web
á Ensures high availability by meeting the Service site (www.comptia.com), for instance, identifies the fol-
Level Agreement requirements, including pro- lowing benefits for prospective CompTIA certified
actively recognizing and responding to problems individuals:
and performing recovery.
á Recognized proof of professional achievement.
á Has thorough working knowledge of hardware A level of competence commonly accepted and
configuration and network connectivity. Includes valued by the industry.
the ability to perform problem determination for
á Enhanced job opportunities. Many employers
all aspects of the server (hardware, software,
give preference in hiring to applicants with certi-
networking).
fication. They view this as proof that a new hire
á Installs and configures server hardware to meet knows the procedures and technologies required.
application requirements.
á Opportunity for advancement. The certification
á Implements current and emerging data storage can be a plus when an employer awards job
and transfer technologies such as SCSI and RAID. advancements and promotions.
á Has a thorough working knowledge of network- á Training requirement. Certification may be
ing protocols (for instance, TCP/IP, IPX/SPX, required as a prerequisite to attending a vendor’s
SNMP, DMI) for diagnosing the impact of the training course, so employers will offer advance
network on the server and vice versa. training to those employees who are already
á Provides support, including second-level support certified.
for resellers and end users. á Customer confidence. As the general public
á Performs maintenance on server systems, data learns about certification, customers will require
storage subsystems, and network devices. that only certified technicians be assigned to their
accounts.
á Has good planning and integration skills to be
able to upgrade a server without impacting net- For any additional information or clarification of the
work users, increase storage capacity without CompTIA Server+ certification path and its history
impacting network users, design and implement a and benefits, consult the CompTIA home page at
data recovery plan in the event of a network www.comptia.com.
device failure; perform peripheral upgrades, BIOS
upgrades, memory upgrades, processor upgrades,
mass storage upgrades, and adapter upgrades.
21 Serv+ Ch_AppC 7/10/01 10:06 AM Page 563
A P P E N D I X C
What’s on the
CD-ROM
This appendix is a brief rundown of what you will find chapter in an electronic version of the book where the
on the CD-ROM that accompanies this book. For a topic is covered. Again, for a complete description of
more detailed description of the newly developed the benefits of ExamGear, see Appendix D.
ExamGear test engine, exclusive to New Riders
Before running the ExamGear software, be sure that
Publishing, see Appendix D, “Using the ExamGear,
AutoRun is enabled. If you prefer not to use AutoRun,
Training Guide Edition Software.”
you can run the application from the CD by clicking
START.EXE.
EXAMGEAR
ExamGear is a test engine developed exclusively for
EXCLUSIVE ELECTRONIC
New Riders Publishing. It is, we believe, the best test
engine available because it closely emulates the format
VERSION OF TEXT
of the Server+ exam. The CD contains 150 questions. The CD-ROM also contains the electronic version of
Each time you run ExamGear, it randomly selects a set this book in Portable Document Format (PDF). In
of questions from the question database, so you never addition to the links to the book that are built in to
take the same exam twice. ExamGear, you can use this version to help search for
In addition to providing a way to evaluate your know- terms you need to study or other book elements. The
ledge of the exam material, ExamGear features several electronic version comes complete with all figures as
innovations that help you to improve your mastery of they appear in the book.
the subject matter. ExamGear also includes an adaptive
testing feature that simulates the adaptive testing method
that CompTIA eventually converts its exams into. COPYRIGHT INFORMATION AND
The practice tests enable you to check your score by
exam area or category, to determine which topics you
DISCLAIMER
need to study further. Other test preparation modes New Riders Publishing’s ExamGear test engine:
provide immediate feedback on your responses, expla- Copyright © 2002 New Riders Publishing. All rights
nations of correct answers, and even hyperlinks to the reserved. Made in U.S.A.
21 Serv+ Ch_AppC 7/10/01 10:06 AM Page 564
22 Serv+ Ch_AppD 7/10/01 10:08 AM Page 565
A P P E N D I X D
Using the ExamGear,
Training Guide
Edition Software
Database
The ExamGear, Training Guide Edition database stores Installing ExamGear, Training
a group of around 150 test questions and answers. The
questions are organized by chapters.
Guide Edition
Install ExamGear, Training Guide Edition by running
the setup program on the ExamGear, Training Guide
INSTALLING AND REGISTERING Edition CD. Follow these instructions to install the
Training Guide Edition on your computer:
EXAMGEAR, TRAINING GUIDE 1. Insert the CD in your CD-ROM drive. The
EDITION AutoRun feature of Windows launches the soft-
ware. If you have AutoRun disabled, click Start
This section provides instructions for ExamGear, and choose Run. Go to the root directory of the
Training Guide Edition installation and describes the CD and choose START.EXE. Click Open and
process and benefits of registering your Training Guide then click OK.
Edition product.
2. Click the button in the circle. A Welcome screen
appears. From here you can install ExamGear.
Requirements Click the ExamGear button to begin installation.
ExamGear requires a computer with the following: 3. The Installation Wizard appears onscreen and
prompts you with instructions to complete the
á Microsoft Windows 95, Windows 98, Windows installation. Select a directory on which to install
NT 4.0, Windows 2000, or Windows ME. ExamGear, Training Guide Edition.
A Pentium or later processor is recommended. 4. The Installation Wizard copies the ExamGear,
á 20–30MB free disk space. Training Guide Edition files to your hard drive,
adds ExamGear, Training Guide Edition to your
á A minimum of 32MB of RAM. Program menu, adds values to your Registry, and
As with any Windows application, the more installs the test engine’s DLLs to the appropriate
memory, the better your performance. system folders. To ensure that the process was
successful, the setup program finishes by running
á A connection to the Internet. ExamGear, Training Guide Edition.
An Internet connection is not required for the
software to work, but it is required for online
registration and to download product updates.
22 Serv+ Ch_AppD 7/10/01 10:08 AM Page 568
5. The Installation Wizard logs the installation you purchased ExamGear, Training Guide Edition from
process and stores this information in a file named a retail store or over the Internet. The information will
INSTALL.LOG. This log file is used by the unin- be used by our software designers and marketing
stall process in the event that you choose to department to provide us with feedback about the
remove ExamGear, Training Guide Edition from usability and usefulness of this product. It takes only a
your computer. Because the ExamGear installation few seconds to fill out and transmit the registration
adds Registry keys and DLL files to your computer, data. A confirmation dialog box appears when registra-
it is important to uninstall the program appropri- tion is complete.
ately. (See the section “Removing ExamGear,
After you have registered and transmitted this informa-
Training Guide Edition from Your Computer.”)
tion to New Riders, the registration option is removed
from the pull-down menus.
It is important that the INSTALL.LOG file be present The Practice Exam section has many of the same
in the directory where you have installed ExamGear, options as Study mode, but you cannot reveal the
Training Guide Edition should you ever choose to answers. This way, you have a more traditional testing
uninstall the product. Do not delete this file. The environment with which to practice.
INSTALL.LOG file is used by the uninstall process to
The Adaptive Exam questions continuously monitor
safely remove the files and Registry settings that were
your expertise in each tested topic area. If you reach a
added to your computer by the installation process.
point at which you either pass or fail, the software ends
the examination. As in the Practice Exam section, you
cannot reveal the answers.
USING EXAMGEAR, TRAINING
GUIDE EDITION
ExamGear is designed to be user friendly and very
intuitive, eliminating the need for you to learn some
confusing piece of software just to practice answering
questions. Because the software has a smooth learning
curve, your time is maximized because you start
practicing almost immediately.
á You can control the order in which questions are Below the Exam drop-down list, you see the questions
presented—random order or in order by skill area that are available for the selected exam. Each exam has
(chapter). at least one question set. You can select the individual
question set or any combination of the question sets if
To begin testing in Study mode, click the Study
more than one is available for the selected exam.
Mode button from the main Interface screen. You
are presented with the Study Mode configuration page Below the Question Set list is a list of skill areas or
(see Figure D.3). chapters on which you can concentrate. These skill areas
or chapters reflect the units of exam objectives defined
At the top of the Study Mode configuration screen, you
by CompTIA for the exam. Within each skill area you
see the Exam drop-down list. This list shows the acti-
will find several exam objectives. You can select a single
vated exam that you have purchased with your
skill area or chapter to focus on, or you can select any
ExamGear, Training Guide Edition product, as well as
combination of the available skill areas/chapters to cus-
any other exams you may have downloaded or any pre-
tomize the exam to your individual needs.
view exams that were shipped with your version of
ExamGear. Select the exam with which you want to In addition to specifying which question sets and skill
practice from the drop-down list. areas you want to test yourself on, you also can define
which questions are included in the test based on your
previous progress working with the test. ExamGear,
Training Guide Edition automatically tracks your
progress with the available questions. When configur-
ing the Study-mode options, you can opt to view all
the questions available within the question sets and skill
areas you have selected, or you can limit the questions
presented. Choose from the following options:
á Select from All Available Questions. This
option causes ExamGear, Training Guide Edition
to present all available questions from the selected
question sets and skill areas.
á Exclude Questions I Have Answered Correctly
X or More Times. ExamGear offers you the
F I G U R E D. 3
The Study Mode configuration page.
option to exclude questions that you have previ-
ously answered correctly. You can specify how
many times you want to answer a question cor-
rectly before ExamGear considers you to have
mastered it. (The default is two times.)
22 Serv+ Ch_AppD 7/10/01 10:08 AM Page 572
á Select Only Questions That I Have Missed X á Order by Skill Area. This option causes
or More Times. This option configures ExamGear to group the questions presented in
ExamGear, Training Guide Edition to drill you the exam by skill area. All questions for each
only on questions that you have missed repeated- selected skill area are presented in succession. The
ly. You may specify how many times you must test progresses from one selected skill area to the
miss a question before ExamGear determines that next, until all the questions from each selected
you have not mastered it. (The default is two skill area have been presented.
times.)
ExamGear offers two options for scoring your exams.
At any time, you can reset ExamGear, Training Guide Select one of the following options:
Edition’s tracking information by clicking the Reset
á Grade at the End of the Test. This option con-
button for the feature you want to clear.
figures ExamGear, Training Guide Edition to
At the top-right side of the Study Mode configuration score your test after you have been presented with
sheet, you can see your access level to the question sets all the selected exam questions. You can reveal
for the selected exam. Access levels are either Full or correct answers to a question, but if you do, that
Preview. For a detailed explanation of each of these question is not scored.
access levels, see the section “Obtaining Updates” in
á Grade as I Answer Each Question. This option
this appendix.
configures ExamGear to grade each question as
Under your access level, you see the score required to you answer it, providing you with instant feed-
pass the selected exam. Below the required score, you back as you take the test. All questions are scored
can select whether the test will be timed and how much unless you click the Show Answer button before
time will be allowed to complete the exam. Select the completing the question.
Stop Test After 90 Minutes check box to set a time
limit for the exam. Enter the number of minutes you You can return to the ExamGear, Training Guide Edition
want to allow for the test. (The default is 90 minutes.) main startup screen from the Study Mode configuration
Deselecting this check box enables you to take an exam screen by clicking the Main Menu button. If you need
with no time limit. assistance configuring the Study-mode exam options,
click the Help button for configuration instructions.
You also can configure the number of questions included
in the exam. The default number of questions changes After you have finished configuring all the exam
with the specific exam you have selected. Enter the options, click the Start Test button to begin the exam.
number of questions you want to include in the exam
in the Select No More than X Questions option.
Starting Practice Exams and
You can configure the order in which ExamGear,
Training Guide Edition presents the exam questions. Adaptive Exams
Select from the following options: This section describes the practice and adaptive exams,
á Display Questions in Random Order. This defines the differences between these exam options and
option is the default option. When selected, it the Study-mode option, and provides instructions for
causes ExamGear, Training Guide Edition to pre- starting them.
sent the questions in random order throughout
the exam.
22 Serv+ Ch_AppD 7/10/01 10:08 AM Page 573
Differences Between the Practice even more difficult. If you answer the question incor-
and Adaptive Exams and rectly, however, the exam “adapts” to your skill level by
presenting you with another question of equal or lesser
Study Mode difficulty on the same subject. If you answer that ques-
Question screens in the practice and adaptive exams are tion correctly, the test begins to increase the difficulty
identical to those found in Study mode, except that the level again. You must correctly answer several questions
Show Answer, Grade Answer, and Item Review buttons at a predetermined difficulty level to pass the exam.
are not available while you are in the process of taking After you have done this successfully, the exam is ended
a practice or adaptive exam. The practice exam provides and scored. If you do not reach the required level of
you with a report screen at the end of the exam. The difficulty within a predetermined time (typically 30
adaptive exam gives you a brief message indicating minutes) the exam is ended and scored.
whether you have passed or failed the exam.
When taking a practice exam, the Item Review screen is
not available until you have answered all the questions.
Why Do Vendors Use
This is consistent with the behavior of most vendors’ Adaptive Exams?
current certification exams. In Study mode, Item
Many vendors who offer technical certifications have
Review is available at any time.
adopted the adaptive testing technique. They have
When the exam timer expires, or if you click the End found that it is an effective way to measure a candi-
Exam button, the Examination Score Report screen date’s mastery of the test material in as little time as
comes up. necessary. This reduces the scheduling demands on the
test taker and allows the testing center to offer more
Starting an Exam tests per test station than they could with longer, more
From the ExamGear, Training Guide Edition main traditional exams. In addition, test security is greater,
menu screen, select the type of exam you want to run. and this increases the validity of the exam process.
Click the Practice Exam or Adaptive Exam button to
begin the corresponding exam type.
Studying for Adaptive Exams
Studying for adaptive exams is no different from study-
What Is an Adaptive Exam? ing for traditional exams. You should make sure that
To make the certification testing process more efficient you have thoroughly covered all the material for each of
and valid and therefore make the certification itself the test objectives specified by the certification exam
more valuable, some vendors in the industry are using a vendor. As with any other exam, when you take an
testing technique called adaptive testing. In an adaptive adaptive exam, either you know the material or you
exam, the exam “adapts” to your abilities by varying the don’t. If you are well prepared, you will be able to pass
difficulty level of the questions presented to you. the exam. ExamGear, Training Guide Edition enables
you to familiarize yourself with the adaptive exam test-
The first question in an adaptive exam is typically an ing technique. This will help eliminate any anxiety you
easy one. If you answer it correctly, you are presented might experience from this testing technique and allow
with a slightly more difficult question. If you answer you to focus on learning the actual exam material.
that question correctly, the next question you see is
22 Serv+ Ch_AppD 7/10/01 10:08 AM Page 574
Multiple Choice
Algorithm Assumptions Most of the questions you see on a certification exam
This section describes the assumptions used in design- are multiple choice (see Figure D.4). This question type
ing the ExamGear, Training Guide Edition Adaptive asks you to select an answer from the list provided.
Exam scoring algorithm. Sometimes you must select only one answer, often indi-
cated by answers preceded by option buttons (round
Chapter Scores selection buttons). At other times, multiple correct
You fail a chapter (and the exam) if any chapter score answers are possible, indicated by check boxes preced-
falls below 66%. ing the possible answer combinations.
22 Serv+ Ch_AppD 7/10/01 10:08 AM Page 575
á Run Simulation. This button is enabled if the á << Previous. This button displays the previous
question supports a simulation. (It is disabled for question on the exam.
this book.) Clicking this button begins the simu-
á Next >>. This button displays the next question
lation process.
on the exam.
á Exhibits. This button is enabled if exhibits are
á << Previous Marked. This button displays if you
provided to support the question. An exhibit is
have opted to review questions that you have
an image, video, sound, or text file that provides
marked using the Item Review screen. This but-
supplemental information needed to answer the
ton displays the previous marked question.
question. If a question has more than one exhibit,
Marking questions is discussed in more detail
a dialog box appears, listing exhibits by name. If
later in this appendix.
only one exhibit exists, the file is opened immedi-
ately when you click the Exhibits button. á << Previous Incomplete. This button displays if
you have opted to review questions that you have
á Reset. This button clears any selections you
not answered using the Item Review screen. This
have made and returns the question window
button displays the previous unanswered question.
to the state in which it appeared when it was
first displayed. á Next Marked >>. This button displays if you
have opted to review questions that you have
á Instructions. This button displays instructions
marked using the Item Review screen. This but-
for interacting with the current question type.
ton displays the next marked question. Marking
á Item Review. This button leaves the question win- questions is discussed in more detail later in this
dow and opens the Item Review screen. For a appendix.
detailed explanation of the Item Review screen, see
á Next Incomplete>>. This button displays if you
the “Item Review” section later in this appendix.
have opted to review questions, using the Item
á Show Answer. This button displays the correct Review screen, that you have not answered. This
answer with an explanation of why it is correct. If button displays the next unanswered question.
you choose this option, the current question will
not be scored.
á Grade Answer. If Grade at the End of the Test is Mark Question and
selected as a configuration option, this button is
disabled. It is enabled when Grade as I Answer Time Remaining
Each Question is selected as a configuration ExamGear provides you with two methods to aid in
option. Clicking this button grades the current dealing with the time limit of the testing process. If you
question immediately. An explanation of the cor- find that you need to skip a question or if you want to
rect answer is provided, just as if the Show check the time remaining to complete the test, use one
Answer button were pressed. The question is of the options discussed in the following sections.
graded, however.
á End Exam. This button ends the exam and dis-
plays the Examination Score Report screen.
22 Serv+ Ch_AppD 7/10/01 10:08 AM Page 577
Just click the column heading you want with the left
mouse button, hold that button down, and move the
mouse into the area directly above the Questions table
(the custom grouping area). Release the left mouse but-
ton to drop the column heading into the custom group-
ing area. To accomplish the custom grouping previously
described, first check the Custom Grouping check box.
Then drag the Marked column heading into the custom
grouping area above the Questions table. Next, drag the
Chapter Name column heading into the custom group-
ing area. You will see the two column headings joined
together by a line that indicates the order of the custom
grouping. Finally, drag the Seq column heading into the
custom grouping area. This heading will be joined to
F I G U R E D. 5
The Questions tab on the Item Review screen. the Chapter Name heading by another line indicating
the direction of the custom grouping.
Notice that each column heading in the custom grouping
appear on the column heading indicating the direction area has an arrow indicating the direction in which items
of the sort (ascending or descending). To change the are sorted under that column heading. You can reverse
direction of the sort, click the column heading again. the direction of the sort on an individual column-heading
The Item Review screen also enables you to create a basis using these arrows. Click the column heading in
custom grouping. This feature enables you to sort the the custom grouping area to change the direction of the
questions based on any combination of criteria you pre- sort for that column heading only. For example, using
fer. For instance, you might want to review the ques- the custom grouping created previously, you can display
tion items sorted first by whether they were marked, the question list sorted first in descending order by
then by the chapter name, and then by sequence num- whether the question was marked, in descending order
ber. The Custom Grouping feature enables you to do by chapter name, and then in ascending order by
this. Start by checking the Custom Grouping check sequence number.
box (see Figure D.6). When you do so, the entire ques- The custom grouping feature of the Item Review screen
tions table shifts down a bit onscreen, and a message gives you enormous flexibility in how you choose to
appears at the top of the table that reads Drag a Column review the exam questions. To remove a custom group-
Header Here to Group by That Column. ing and return the Item Review display to its default
setting (sorted in descending order by sequence num-
ber), just uncheck the Custom Grouping check box.
22 Serv+ Ch_AppD 7/10/01 10:08 AM Page 579
Examination Score
Report Screen
F I G U R E D. 7 The Examination Score Report screen appears when
The Current Score tab on the item review screen.
the study Mode, practice exam, or adaptive exam
ends—as the result of timer expiration, completion of
all questions, or your decision to terminate early (see
Review Marked Items Figure D.8).
The Item Review screen enables you to enter a browse
This screen provides you with a graphical display of
sequence for marked questions. When you click the
your test score, along with a tabular breakdown of
Review Marked button, questions that you have previ-
scores by chapter. The graphical display at the top of
ously marked using the Mark Question check box are
the screen compares your overall score with the score
presented for your review. While browsing the marked
required to pass the exam. Buttons below the graphical
questions, you will see the following changes to the
display enable you to open the Show Me What I
buttons available:
Missed browse sequence, print the screen, or return to
á The caption of the Next button becomes Next the main menu.
Marked.
á The caption of the Previous button becomes Show Me What I Missed
Previous Marked. Browse Sequence
The Show Me What I Missed browse sequence is
invoked by clicking the Show Me What I Missed but-
Review Incomplete ton from the Examination Score Report or from the
The Item Review screen enables you to enter a browse configuration screen of an adaptive exam.
sequence for incomplete questions. When you click the Note that the window caption is modified to indicate
Review Incomplete button, the questions you did not that you are in the Show Me What I Missed Browse
Sequence mode. Question IDs and position within the
browse sequence appear at the top of the screen, in
place of the Mark Question and Time Remaining indi-
cators. Main window contents vary, depending on the
22 Serv+ Ch_AppD 7/10/01 10:08 AM Page 581
F I G U R E D. 8
The Examination Score Report screen. Frequently Asked Questions
ExamGear FAQ can be found at
http://www.newriders.com/examgear/support/faq.cfm
question type. The following list describes the buttons
available within the Show Me What I Missed browse
sequence and the functions they perform:
OBTAINING UPDATES
á Return to Score Report. Returns control to the
Examination Score Report screen. In the case of This section summarizes procedures for
an adaptive exam, this button’s caption is Exit, obtaining updates.
and control returns to the adaptive exam configu-
ration screen.
á Run Simulation. Opens a simulation in Grade The Catalog Web Site
mode, causing the simulation to open displaying for Updates
your response and the correct answer. If the cur-
rent question does not offer a simulation, this Selecting the Check for Product Updates option from
button is disabled. the Online menu shows you the full range of products
you can either download for free or purchase from your
á Exhibits. Opens the Exhibits window. This but- web browser. You must be connected to the Internet to
ton is enabled if one or more exhibits are avail- reach these web sites.
able for the question.
á Instructions. Shows how to answer the current Types of Updates
question type. Several types of updates may be available for download,
including various free updates and additional items
available for purchase.
22 Serv+ Ch_AppD 7/10/01 10:08 AM Page 582
Free Program Updates You can also reach us on the World Wide Web:
Free program updates include changes to the
http://www.newriders.com
ExamGear, Training Guide Edition executables and
runtime libraries (DLLs). When any of these items are
downloaded, ExamGear automatically installs the Technical Support
upgrades. ExamGear, Training Guide Edition will be
reopened after the installation is complete. Technical support is available at the following phone
number during the hours specified:
Free Database Updates Telephone: 317-581-3833
Free database updates include updates to the exam(s)
that you have registered. Exam updates are contained in Email: examgear@newriders.com
compressed, encrypted files and include exam databases Monday through Friday, 10 a.m.–3 p.m.
and exhibits. ExamGear, Training Guide Edition auto- Central Standard Time.
matically decompresses these files to their proper loca-
tion and updates the ExamGear software to record You can visit the online support web site at
version changes and import new question sets. www.newriders.com/support and submit a support
request over the Internet.
Index
SYMBOLS Windows 2000, 253-254
Windows NT, 253
addresses, IP (Internet Protocol), 31
8mm tape, 85, 506 adequate cooling, verifying, 152
10 Gig Ethernet, 73 Advanced Attributes dialog box, 289
10BASE2, 69 Advanced Intelligent Tape (AIT), 85-86, 472, 506
10BASE5, 69 agents, SNMP (Simple Network Management Protocol), 301-306
10BASEF, 70 AGP (Accelerated Graphics Port), 51
10BASET, 69-70, 175 air quality, 332
100BASEF, 72 AIT (Advanced Intelligent Tape), 85-86, 472, 506
100BASET, 72, 175 American National Standards Institute (ANSI), 58
1000BASELX, 73 amount of memory, 53
1000BASESX, 73 analysis, risk, 453-455
1000BASET, 73 anti-theft devices, 344-346
antivirus applications, 117, 223-224
APIPA (Automatic Private IP Addressing), 388
A AppleTalk, 78
appliances, servers, 26
Application log, Windows NT, 370
Accelerated Graphics Port (AGP), 51
applications
access. See also physical security
antivirus, 117, 223-224
remote-control remote access techniques, 35
backups, 223
remote-node remote access techniques, 35
configuration, 115
removable media, 346-347
diagnostic. See diagnostic tools
access points (APs), 74
ExamGear, 521
Active Server Pages (ASP), 40
performance-monitoring, 422-427
active study strategies, 520
RAID, 461-466
adapters
searching, 244
adding, 267
server roles, 32
BIOS/firmware, 266
system monitor agents, 224-225
configuring, 268
troubleshooting, 438-439
replacing, 266
vendor-specific management, 115-116
SCSI, 159
Applications log, Windows 2000, 384
server management, 181
applying troubleshooter senses, 363
teaming, 74
architecture
upgrading, 265-267
buses, 48-49
video, 160, 181
SSA (Serial Storage Architecture), 65
adding
archive bits, 288
adapters, 267
arrays
hard drives, 256-259
RAID 5, 204-205
memory, 260-263
SCSI RAID, 259
processors, 248-255
tape, 88
SCSI hard disk drives, 258
ASP (Active Server Pages (ASP), 40
additional memory, installing, 262-263
assigning server roles, 26-28, 30-41
additional processors
asymmetric multiprocessing, 46
installing, 249-253
ATA (AT Attachment), 58-59
operating systems, 253
serial ATA, 60-61
Linux, 255
subsystems, 433
Novell NetWare, 254
upgrading, 256-257
1087-2 index 7/11/01 10:51 AM Page 588
588 INDEX
INDEX 589
590 INDEX
INDEX 591
592 INDEX
INDEX 593
594 INDEX
INDEX 595
swap files, 227 Management Information Base (MIB), 80, 223, 303
system shutdown, 413-414 matching
vmstat command, 425 backup types, 290-291
load balance, 268, 301 memory, 165
local area network (LAN), 28 processors, 162-163
local buses, 49-50 MBps (megabytes per second), 59
locks, 342. See also security ME (Millennium Edition), 23
Logical Unit Number (LUN), 218 media backups, 291-296. See also backups
logs megabytes per second (MBps), 59
events, 225 megahertz (MHz), 47
files, 299 memory, 52, 156, 480
interpreting, 416-417 adapters, 265-266
NetWare, 400 adding, 260-263
upgrading, 247 amount of, 53
loops, Fibre Channel Arbitrated Loop disk system, 175 bottlenecks, 428-429
Low-Voltage Differential (LVD), 62, 170, 502 buses, 47
LTO (Linear Tape-Open), 87 error checking, 56
LUN (logical Unit Number), 218 installing, 164
LVD (Low-Voltage Differential), 62, 170, 502 interleaving, 53
speed, 165
types, 53
M form factors, 54-55
matching, 165
technologies, 55-57
Macintosh, AppleTalk, 78 virtual, 53
macro study strategies, 520 Memory-In-Cassette (MIC), 85, 506
magic packets, 75 messages, error, 416-417
mail server roles, 33 Metcalfe, Bob, 68
maintenance, 514-515. See also troubleshooting MHz (megahertz), 47
networks, 286 MIB (Management Information Base), 80, 223, 303
baselines, 297-300 MIC (Memory-In-Cassette), 85, 506
multiple tape sets, 292-293 Micro Channel Architecture, 48
operating systems, 294-297 micro study strategies, 520
performing regular backups, 286-287 Microsoft Management Console (MMC), 382
routines, 293 Millennium Edition (ME), 23
types of backups, 288-291 minix, 212
operating parameters, 325-327 mirrored pairs, 457
air quality, 332 mirrored striped arrays, 459
ESD (electrostatic discharge), 331, 337 mirroring disks, 203-204
fire suppression, 335-336 misconfigurations, troubleshooting, 431-438. See also
flooding, 336 configuration
humidity, 331 mixing backup types, 290-291. See also backups
monitoring server states, 338 MMC (Microsoft Management Console), 382
power lines, 332-334 models
temperature, 328-332 dedicated application, 32
physical environments, 307 distributed application, 33
physical security, 339 peer-to-peer application, 33
anti-theft devices, 344-346 modems, 106, 160
computer data centers, 340-344 MONITOR.NLM (Novell NetWare), 400-401
removable media, 346-347 monitors, 157
Mammoth drives, 472 installing, 179
man command (UNIX/Linux), 423 racking servers, 109
management server states, 338
cables, 108 SNMP agents, 305-306
networks, 79-81, 114-118 system performance, 298-300
SNMP (Simple Network Management Protocol), 222-223, video, 104
301-302
1087-2 index 7/11/01 10:51 AM Page 596
596 INDEX
INDEX 597
protocols, 77-79
Token Ring, 73-74 O
wireless, 74
TCP/IP (Transmission Control Protocol/Internet Protocol), objectives
434-438 adapters, 265-267
troubleshooting, 434, 515-517 applications, 438-439
UNIX, 402-406 backups
upgrading, 243-247 hardware, 82-88
web sites, 419 performing regularly, 286
Windows 2000 strategies, 94-95
ipconfig command, 385-386 baselines, 297-300
nbtstat command, 387-388 BIOS (Basic Input Output System), 198-200, 263-265
netdiag command, 388-391 bottlenecks
netstat command, 392-393 disk subsystems, 429-430
nslookup command, 394-395 memory, 428-429
pathping command, 398-399 network subsystems, 430
ping command, 395-397 processors, 427-428
tracert command, 397-398 troubleshooting, 422-427
Windows NT buses, 46-52
diagnostic tools, 371 diagnostic tools, 364-372, 376-379
ipconfig command, 371-372 disaster recovery, 103, 452
nbtstat command, 372-373 disk drives, 88
netstat command, 374-375 disk subsystems, 58
nslookup command, 376-377 EIDE, 59
ping command, 377-378 Fibre Channel, 65
tracert command, 378-379 IDE/ATA, 58-59
newsgroups NAS, 66-68
IBM OS/2 Warp Server, 418 S.M.A.R.T., 64-65
Linux, 419 SAN, 66
Novell NetWare, 419 SCSI, 62, 64
UNIX, 419 serial ATA, 60-61
Windows NT/2000, 418-419 SSA, 65
NFS (Network File System), 66 Ultra ATA/Ultra DMA, 59
NIC (network interface cards), 176, 481 drivers, 220
fail overs, 92 external peripherals, 216-219
redundant, 182 fault tolerance, 95
to hub cabling, 177 RAID, 96-99
NOS (Network Operating System) UPS, 96
configuring, 215 full system backups, 247. See also backups
installing, 207-215 hard drives, 256-259
updating, 220 hardware, 104-106
verifying connectivity, 216 installing, 161
notification, establishing, 313-314 troubleshooting, 438-439
Novell NetWare. See NetWare verifying, 308
Novell Storage Services (NSS), 210 hardware-based RAID
Novell’s Directory Service (NDS), 111 configuring, 200-202
nslookup command RAID 0 configuration, 203
Windows 2000, 394-395 RAID 0/1 configuration, 205-207
Windows NT, 376-377 RAID 1 configuration, 203-205
NSS (Novell Storage Services), 210 RAID 5 configuration, 204-205
NT File System (NTFS), 111, 208 high availability, 94
NTBACKUP.EXE, 294 Intel processors, 41
NTFS (NT File System), 111, 208 caches, 44-45
NWFS (NetWare File System), 111, 210 Itanium, 44
NWLink, 78 Pentium 4, 43
Pentium II, 42
1087-2 index 7/11/01 10:51 AM Page 598
598 INDEX
INDEX 599
600 INDEX
INDEX 601
602 INDEX
INDEX 603
604 INDEX
tar (tape archive) utility, 296-297 Transaction Tracking System (TSS), 400
target files, backups, 296 Transmission Control Protocol/Internet Protocol (TCP/IP),
Task Manager 77, 505
Windows 2000, 379-381 traps, 302
Windows NT, 365-366 Travan Cartridge Tape, 83-84
TCP/IP (Transmission Control Protocol/Internet Protocol), 77 troubleshooting
configuring, 215 applications, 438-439
troubleshooting, 434-438 ATA subsystems, 433
tcpcfg command (IBM OS/2 Warp Server), 407 boot failure, 438
TCPCON command (NetWare), 401 bottlenecks, 422-430
technologies, memory, 55-57 diagnostic tools. See diagnostic tools
temperature, 328-332 disaster recovery, 452
terminal server roles, 36 backing up data, 471-476
testing. See also exams creating plans, 452-468
disaster recovery plans, 477-479 fault tolerance/recovery, 470-471
existing memory, 260 hot swapping, 468-469
power-on tests, 184 restoring servers, 480-481
pre-testing, 521 testing, 477-479
RAID controller BIOS, 200 disk drives, 431
SCSI controller BIOS, 199-200 existing memory, 260
system BIOS, 199 hardware, 438-439
upgrades, 245 misconfigurations, 431-438
thermometers, 363 networks, 434
thinnet, 69 baselines, 297-300
thresholds multiple tape sets, 292-293
agents, 305-306 operating systems, 294-297
configuring, 300-302 proactive maintenance, 286-287
MIB (Management Information Base), 303 routines, 293
security parameters, 305 servers, 515-517
tracking traffic, 306-307 types of backups, 288-291
time to take exams, 522 operating parameters, 325-327
tips air quality, 332
exam sessions, 522-523 ESD (electrostatic discharge), 331, 337
study, 520-522 fire suppression, 335-336
Token Rings, 73-74, 176 flooding, 336
tools humidity, 331
backups, 288-291 monitoring server states, 338
diagnostic. See diagnostic tools power lines, 332-334
Linux, 296-297 temperature, 328-332
Performance, 298-300 physical security, 339
RDISK.EXE, 295 anti-theft devices, 344-346
server support, 271 computer data centers, 340-344
service, 270-271 removable media, 346-347
System Monitor, 299 problem determination, 361-363
Windows 2000, 295-296 RAID controller BIOS, 200
Windows NT Hardware Query, 313 SCSI (Small Computer System Interface)
Torvalds, Linus, 113 controller BIOS, 199-200
traceroute command subsystems, 432
IBM OS/2 Warp Server, 409 system BIOS, 199
Linux, 406 TCP/IP, 434-438
UNIX, 406 Ultra ATA subsystems, 434
Windows 2000, 397-398 upgrades, 431-438
Windows NT, 378-379 viruses, 438-439
trackballs, racking servers, 109 TSS (Transaction Tracking System), 400
tracking SNMP (Simple Network Management Protocol), 306-307
traffic, SNMP (Simple Network Management Protocol), 306-307
1087-2 index 7/11/01 10:51 AM Page 605
INDEX 605
606 INDEX
H O W TO C O N TA C T U S
EMAIL US
Contact us at: nrfeedback@newriders.com
• If you have comments or questions about this book
• To report errors that you have found in this book
M AT T E R
• If you have a book proposal to submit or are interested in writing for New Riders
• If you are an expert in a computer topic or technology and are interested in being a technical editor
who reviews manuscripts for technical accuracy
Contact us at: nreducation@newriders.com
• If you are an instructor from an educational institution who wants to preview New Riders books for
classroom use. Email should include your name, title, school, department, address, phone number, office
days/hours, text in use, and enrollment, along with your request for desk/examination copies and/or
additional information.
Contact us at: nrmedia@newriders.com
• If you are a member of the media who is interested in reviewing copies of New Riders books. Send
your name, mailing address, and email address, along with the name of the publication or web site you
work for.
T H AT
B U L K P U R C H A S E S / C O R P O R AT E S A L E S
If you are interested in buying 10 or more copies of a title or want to set up an account for your com-
pany to purchase directly from the publisher at a substantial discount, contact us at 800-382-3419 or
email your contact information to corpsales@pearsontechgroup.com. A sales representative will contact
you with more information.
W R I T E TO U S
New Riders Publishing
V O I C E S
C A L L / FA X U S
Toll-free (800) 571-5840
If outside U.S. (317) 581-3500
Ask for New Riders
F A X : (317) 581-4663
W W W. N E W R I D E R S . CO M
Server+ BM 7/11/01 8:38 AM Page 2
Core Exams
New MCSE Candidates (Who Have Not Already Passed Windows NT 4.0 Exams) MCPs Who Have Passed 3 Windows NT 4.0
Must Take All 4 of the Following Core Exams: Exams (Exams 70-067, 70-068, and 70-073)
ISBN 0-7357-0988-2
(Selected third-party certifications that focus on interoperability will be accepted as an alternative to one
elective exam. Please watch for more information on the third-party certifications that will be acceptable.)
*Exam 70-219: Designing a Microsoft Windows 2000 Directory Services Infrastructure
*Exam 70-220: Designing Security for a Microsoft Windows 2000 Network
*Exam 70-221: Designing a Microsoft Windows 2000 Network Infrastructure ISBN 0-7357-0983-1 ISBN 0-7357-0984-X ISBN 0-7357-0982-3
Exam 70-222: Upgrading from Microsoft Windows NT 4.0 to Microsoft Windows 2000
*Core exams that can also be used as elective exams may only be counted once toward a certification; that is, if a candidate receives
credit for an exam as a core in one track, that candidate will not receive credit for that same exam as an elective in that same track.
WWW.NEWRIDERS.COM
Server+ BM 7/11/01 8:38 AM Page 3
R E L AT E D N E W R I D E R S T I T L E S
When
IT really
matters,
test with
You’ve studied the Training Guide. Tested your skills with ExamGear ™.
Now what? Are you ready to sit the exam?
If the answer is yes, be sure to test with VUE.
Why VUE? Because with VUE, you get the best technology and even better
service. Some of the benefits are:
VUE is a testing vendor for all the major certification vendors, including Cisco®,
Microsoft®, CompTIA® and Novell®. Coming soon, you’ll find New Riders
questions and content on the VUE web site, and you’ll be able to get
your next Training Guide at www.vue.com.
www.newriders.com W W W . V U E . C O M
Server+ BM 7/11/01 8:38 AM Page 5
www.informit.com ■ www.newriders.com
Server+ BM 7/11/01 8:38 AM Page 6