VMW WP 10.1

VMware vSphere with 3PAR Utility Storage
Simple and Efficient VMware vSphere Deployment with 3PAR InServ Storage Servers
Maneesh Jain Storage Architect
Document Abstract:
This paper discusses the benefits of deploying VMware vSphere with 3PAR InServ Storage Servers for greater consolidation, cost savings, and simplified administration in virtualized server environments, and lays out best practices for an integrated VMware vSphere plus 3PAR Utility Storage deployment. Part Number: vmw-wp-10.1

-2-
Table of Contents
1 2 3 Introduction ............................................................................................................................... 3 Overview ................................................................................................................................... 3 Greater VM Consolidation ....................................................................................................... 4 3.1 Performance and Reliability Features ............................................................................ 5 3.2 Queue Depth Throttling ................................................................................................... 5 3.3 Metadata Locking with SCSI Reservations.................................................................... 6 3.4 VMware vStorage APIs for Array Integration (VAAI) ..................................................... 7 3.5 Storage I/O Controls ........................................................................................................ 8 3.6 VMFS versus RDM ......................................................................................................... 8 4 Cost Savings ............................................................................................................................ 8 4.1 Virtual Volumes: Thin or Fat? ......................................................................................... 8 4.2 Virtual Disk Format Options .......................................................................................... 10 4.3 3PAR Thin Conversion and Thin Persistence ............................................................. 10 5 Increased Administrative Efficiency ...................................................................................... 13 5.1 3PAR Plug-In for VMware vCenter............................................................................... 13 5.2 3PAR Recovery Manager for VMware vSphere.......................................................... 14 5.3 3PAR Replication Adapter for VMware vCenter Site Recovery Manager (SRM) ..... 15 5.4 3PAR Dynamic Optimization and VMware vSphere ................................................... 15 5.5 Boot-from-SAN Environments ...................................................................................... 16 6 Conclusion .............................................................................................................................. 17
Appendix: Frequently Asked Questions ...................................................................................... 18
2010 3PAR Inc. All rights reserved. 3PAR, the 3PAR logo, Serving Information, InServ, InForm, InSpire, and Thin Built In are all trademarks or registered trademarks of 3PAR Inc. All other trademarks and registered trademarks are the property of their respective owners.
-3-
Introduction
Server virtualization from VMware provides a powerful environment for consolidating a large number of servers, thereby delivering cost savings and flexibility in the datacenter. 3PAR Utility Storage provides an ideal complement to VMware vSphere deployments by delivering a virtualized storage platform to support server virtualization. This paper presents the benefits of using 3PAR InServ Storage Servers for server virtualization with VMware vSphere and documents best practices for this type of deployment.
Overview
When architected with the correct underlying storage, server virtualization delivers higher cost savings, greater consolidation, and increased administrative efficiency that are not only transforming the datacenter but also the businesses that those datacenters fuel. However, these transformative results depend on enterprise storage to deliver the performance, availability, and flexibility to keep up with the dynamic and consolidated nature of virtualized server environments. The 3PAR Utility Storage platform exceeds the requirements placed on storage infrastructures by server virtualization. When deployed together, VMware vSphere and 3PAR Utility Storage deliver a compelling utility computing solution that combines server virtualization with highly virtualized storage for combinatory benefit. Complementary features inherent to both platforms increase overall utilization, provisioning agility, and administrative efficiency. In addition, 3PAR Utility Storage uniquely enhances the flexibility and return on investment (ROI) of VMware vSphere deployments in the following ways: Greater virtual machine (VM) consolidation. The unique 3PAR InSpire Architecture stripes volumes widely across all disks to deliver maximum I/O throughput and minimum latencies, which mitigates server memory and traditional storage constraints. Increased array performance can not only boost VM-based application performance, but when paired with the superior reliability of the InServ platform and advanced support of VMwares VAAI capabilities, higher VM density and therefore greater VM consolidation are possible. This benefit can reduce the number of physical servers required by VMware vSphere environments by up to 50%. Simplified storage administration. 3PAR Utility Storage reduces storage administration time by up to 90%. Autonomic management features such as 3PAR Autonomic Groups automate repetitive storage administration tasks while autonomic storage tiering tools like 3PAR Dynamic Optimization allow administrators to tune storage allocations without interruption to servers. Integration with the VMware vCenter Server management console. The 3PAR Plug-In for VMware vCenter allows administrators to monitor 3PAR volumes from within the vSphere console. 3PAR Recovery Manager for VMware vSphere uses 3PAR Virtual Copy to create point-in-time, disk-based snapshots, giving VMware administrators a simple process for recovering whole Virtual Machine Disks (VMDKs), individual VMs, or even individual files. Simplified disaster recovery. 3PAR Replication Adapter for VMware vCenter Site Recovery Manager (SRM) provides simple and reliable replication and disaster recovery for VMware vCenter environments based on 3PAR Remote Copy.
-4-
Broader VMware vSphere deployment through cost-effective 3PAR Thin Provisioning. With 3PAR, you can add more flexibility to your VMware environment and reduce physical storage capacity needs by leveraging the cost-effective usable capacity delivered by Fast RAID 5 and RAID 6 (also called RAID Multi-Parity or RAID MP) and 3PAR Thin Provisioning, which allows customers to safely overallocate capacity up front, and then purchase actual physical capacity incrementally and only for written data. The result can be a savings of 50% or more in purchased capacity for VMware vSphere deployments. 3PAR support for VMware vClient initiatives. 3PAR InServ arrays provide rapid provisioning of both storage and virtual desktops for VMware View and VMware View Composer deployments. Together, these solutions provide storage efficiency for boot images as well as user data, and deliver the performance, scalability, and high availability that an enterprise vClient deployment demands.
Greater VM Consolidation
Consolidating workloads in virtual server environments is key to driving increased savings. The number of virtual machines per physical serverknown as VM densityis a leading indicator of consolidation success and is a function of higher performance and reliability. 3PARs wide striping, mixed workload support, and integration with the VMware Adaptive Queue Depth Throttling algorithm provide the performance necessary to increase the VM density per server by at least twofold when compared to virtual server deployments using traditional storage arrays. Since a VMware ESX server can host many different VMs, each with its own I/O patterns, ESX I/O is generally random in nature. Random I/O means the ESX server derives less benefit from storage caching than a server with a more predictable access pattern, forcing more requests to access disk. Thus, performance is determined by the number of disks (IOPS density) that make up the LUN that is presented to the ESX server. 3PARs wide striping spreads even modestly sized volumes across all disks in the array, resulting in unmatched I/O performance even when caching is minimally effective. In addition to using storage for files and similar data, VMware ESX implements aggressive memory management techniques, including higher rates of paging and swapping to/from disk than are typical of contemporary, non-virtualized servers. VMware therefore recommends placing swap files on high-performance storage. In ESX deployments, the amount of memory that can be installed in the server often limits the number of virtual machines that can be placed on a single physical server. 3PAR customers have found that the improved swapfile performance afforded by using 3PAR InServ arrays allows a twofold to fourfold increase in the number of VMs per physical server, which is otherwise constrained by server memory. The Increased reliability of 3PAR InServ arrays is a function of architectural features such as redundant hardware, fast RAID rebuild capability, and RAID isolation as well as InForm Operating System features such as 3PAR Persistent Cache. These features collectively minimize service outages, allowing greater consolidation in virtualized server environments without undue risk. The following sections provide an overview of 3PAR features which contribute to higher performance and reliability in a VMware environment, and then detail a number of best practices for maximizing performance with ESX and 3PAR.
-5-
3.1
Performance and Reliability Features
As mentioned previously, the 3PAR InSpire Architecture includes a number of features that drive higher levels of performance and increase reliability across a VMware vSphere deployment: Wide striping distributes each virtual volume (seen as a LUN by an ESX server) across all disks and other resources of the array. This applies to both VMware vStorage Virtual Machine File System (VMFS) and Raw Device Mapping (RDM) volumes, including those which use 3PAR Thin Provisioning software. Wide striping allows even small volumes to gain the performance benefits of many disks, and provides for consistent levels of performance even at increased levels of capacity utilization. Mesh-Active cluster controllers make all array resources available to all LUNs at all times, which provides the host with symmetric, active-active access to all LUNs. Unlike traditional arrays, where controllers have preferred disks and access to foreign disks within the array is expensive, 3PARs high-performance, Mesh-Active backplane enables each LUN to have uniformly high performance access to all resources within the arrayregardless of which controller nodes are used to connect an ESX server to the array. ESX servers gain the full performance benefits of the Mesh-Active design by changing from the default "Fixed" pathing to "Round Robin" multipathing. Adaptive Cache provides dynamic and autonomic balancing of read and write cache levels to increase performance by adjusting caching without the need for manually setting or modifying policies or schedules. Mixed workload support enables different types of applications (involving both transactionbased and throughput-intensive workloads) to run without contention on a single InServ array. Mixed workload support is especially important in consolidated and virtualized environments, where the same array must support a wide mix of application types across the cluster. Persistent Cache provides resiliency in the event of a node failure by using the highperformance, Mesh-Active backplane to quickly re-mirror write-back cache to other nodes in the system. In the event of a node failure, Persistent Cache enables the InServ array to maintain the substantial performance benefits of write-back caching without risk of data loss. This feature allows always on application and virtual server environments to gracefully handle an unplanned controller failure without the substantial performance penalties associated with traditional arrays and write-through mode. RAID isolation preserves access to data even if an entire disk chassis (up to 40 disks) is lost. Traditional RAID merely guarantees continued data access in the event of a single disk, or two disks with RAID 6.
3.2
Queue Depth Throttling
Each 3PAR port has a finite queue depth that depends on the HBA model (see Table 1) and each ESX server attached via a port shares that connection ports queue. If an ESX host sends an I/O request to a port with a full queue, the host receives a queue full SCSI response from the InServ array. Historically, an ESX hosts default reaction to this response would be to recognize it as a valid command and to continue sending requests to that port. I/O commands sent to a 3PAR port that has reached its maximum queue depth are not processed beyond the queue full SCSI response. Lack of I/O response results in VMs becoming unresponsive and can lead to a crash of the ESX server.
-6-
This behavior to a queue full SCSI condition is addressed by VMware as of ESX 3.5 Update 4 and is included in the latest version of ESX. The remedy consists of an adaptive queue depth throttling algorithm introduced by VMware that adjusts the LUN queue depth in the VMkernel I/O stack. This algorithm is activated when the storage array indicates I/O congestion by returning a queue full SCSI status. When congestion is detected, VMkernel throttles the LUN queue depth and attempts to gradually restore the queue depth when congestion conditions subside. Without adaptive queue depth throttling, administrators are forced to limit the number of VMs per physical server so as to reduce the risk associated with any particular VM overrunning I/O queues. Administrators are also forced to manually tune the number of VMs when they detect congestiona reactive, slow, and error-prone process. By automating congestion control, administrators can confidently create a higher number of VMs per physical server without the need for manual congestion control. For more information, refer to this KB entry on how to enable the VMware adaptive queue depth throttling algorithm: http://kb.vmware.com/selfservice/documentLink.do?externalID=1008113. If it is not possible to enable VMwares adaptive queue depth throttling algorithm, then follow the recommendations in the 3PAR VMware ESX 3.X 4.X Implementation Guide to tune/throttle the ESX host.
HBA Type
QLogic LSI Logic Emulex 3PAR
Speed
2 Gbps 2 Gbps 4 Gbps 4 Gbps
Queue Depth
497 510 959 1638
Table 1. Queue Depth for 3PAR Fibre Channel HBAs Tech Tip: With either the VMware adaptive queue depth algorithm or target throttling on the ESX host, no more than 16 ESX hosts should be attached to any one 3PAR port rated at 2 Gbps. For 4-Gbps ports, the recommended limit is 32 ESX hosts. It is also important to pay attention to the number of VMs that are hosted on each ESX server. For example, if you have four ESX servers connected to a given 3PAR port, and each of those four ESX servers has 20 VMs, this is the equivalent of 80 servers going into one 3PAR port. You would want to monitor the Qlen values on the InServ (using the command statvlun ni rw host <ESX host>) to make sure you are not exceeding these values.
3.3
Metadata Locking with SCSI Reservations
Metadata updates by ESX must ensure exclusive access when multiple ESX hosts are accessing shared storage. Locking is necessary to prevent multiple hosts from concurrently writing to the metadata, which would lead to data corruption. Examples of operations that require metadata updates include: powering on/off a VM, adding/removing a VMFS datastore, creating a new VM, using VMware VMotion, and growing a VMFS datastore. Prior to vSphere 4.1, SCSI reservations were used to perform this locking. With vSphere 4.1 and later, what VMware refers to as Hardware Assisted Lockingmade possible with the ATS SCSI
-7-
commandcan be used. The following section details the ATS command and other improvements introduced with vSphere 4.1 in conjunction with the VAAI initiative, which is also described below. Prior to Hardware Assisted Locking, a SCSI reservation (basically a lock on the LUN) was made before metadata was ready to be updated, and the lock was released after the metadata update operation had completed, allowing other operations to continue. Under this system, too many reservations being made at once could lead to I/O failures if a host was unable to make a reservation because another host had a lock on the LUN. When a host is unable to make a reservation due to a conflict with another host, it retries at random intervals until it is successful. If too many unsuccessful retries are made, the operation fails. With SCSI reservations, it is best to limit the number of operations that can cause reservation conflicts and stagger them so that relatively few reservations are attempted at the same time. For example, it is not advisable to power on/off multiple VMs at the same time or to run multiple instances of VMotion in parallel. However, as mentioned above, these challenges and exclusive LUN locks have been addressed with the latest vSphere release, as discussed below.
3.4
VMware vStorage APIs for Array Integration (VAAI)
VMware vStorage is a set of technologies and interfaces that enable vSphere to leverage storage resources to deliver the efficiency, control, and ease of customization that customers demand of their IT environment. VAAI (vStorage APIs for Array Integration) is one of these vStorage technologies; under the VAAI initiative, APIs have been introduced to improve performance and scalability by leveraging efficient array-based operations. 3PAR has developed the 3PAR Plug-In for VMware VAAI to deliver enhanced performance, agility, and scalability using the following vSphere commands introduced in 4.1: Hardware Assisted Locking (ATS) eliminates SCSI reservation contention by providing a fast, fine-grained locking mechanism. The ATS (Atomic Test and Set) command verifies that a block of metadata is what is expected (test) and then replaces it with an updated block (set) in a single, atomic operation. Using this command, the ESX server can lock a portion of a LUN related to a single VM instead of locking the whole LUN as described in the previous section, thereby allowing other VMs on the same LUN to continue operating normally. 3PARs implementation of ATS uses the 3PAR Gen3 ASIC to further improve performance. The combination of ATS and the 3PAR Gen3 ASIC allows an increase in VM density per LUN and greater scalability for vSphere deployments. Fast Copy (XCOPY) improves the performance of common operations like virtual machine cloning and Storage vMotion by performing large data movement operations directly within the storage array. By not requiring each block to make a round-trip to the host, the time required for these operations is significantly reduced while storage network traffic is minimized. When combined with 3PAR Thin Persistence, disk I/O and storage capacity can also be reduced since blocks of zeros are not written due to the arrays zero-detection capability, which is integrated into the Gen3 ASIC. Block Zeroing (WRITE_SAME) uses the standard SCSI command WRITE_SAME to offload large, block-level write operations of zeros from the host to the storage array. Block zeroing improves host performance and efficiency when allocating or extending Eager Zeroed Thick (EZT) virtual disks, or on initial access to a block on a non-EZT virtual disk. When combined with 3PARs built-in zero-detection and EZT virtual disks,
-8-
storage array bandwidth, disk I/O bandwidth, and disk consumption is minimized. Initialization of EZT virtual disks in seconds rather than minutes eliminates the tradeoff between fast VM creation and fast run-time performance. Using these vSphere 4.1 enhancements, 3PAR storage arrays provide an even more VM-dense and performance-efficient environment than with earlier versions of vSphere.
3.5
Storage I/O Controls
Storage I/O Controls (SIOC), added in vSphere 4.1, is a feature which protects the storage performance of high-priority virtual machines. It does this by detecting high storage I/O latency (in excess of 20 ms response time). Once excess latency is detected, SIOC throttles the host-side queue depths of lower-priority virtual machines (based on administrator-defined shares of storage I/O bandwidth) to prevent low-priority VMs from gaining more than their share of I/O bandwidth.
3.6
VMFS versus RDM
VMware Virtual Machine File System (VMFS) is a clustered, hierarchical file system that is designed and optimized by VMware specifically for virtualized server environments. It is the container for the virtual disks used by virtual machines (VMs). VMFS increases resource utilization by allowing multiple ESX servers to access shared storage concurrently. It provides distributed locking for VMs so all VMs can operate safely in a Storage Area Network (SAN). In addition, VMFS uses an optimized architecture to enable users to achieve higher VM-to-physicalmachine server density. VMs are represented as a Virtual Machine Disk (VMDK) in a VMFS datastore. Raw Device Mapping (RDM) is an alternative to VMFS. RDMs provide VMs with direct I/O access to raw storage volumes. RDMs are used for the quorum disks in clusters such as Microsoft Cluster Server (MSCS) VMs.
Cost Savings
Todays IT administrator faces potentially spiraling storage costs as a result of the explosive growth of data and the new storage performance challenges created by technologies such as server virtualization. 3PARs massively parallel architecture uniquely addresses these performance challenges while 3PARs Thin Technologiesincluding 3PAR Thin Provisioning, 3PAR Thin Conversion, and 3PAR Thin Persistenceenable organizations to meet storage needs with 50% less capacity, energy, and floor space.
4.1
Virtual Volumes: Thin or Fat?
VMware recommends that RAID groups be grown by adding disks to the volumes that are 1 presented to the ESX server in order to satisfy the ESX I/O requirements. LUNs presented to ESX servers should be striped across many physical disks. VMware also recommends that a large LUN be presented to ESX as a VMFS datastore for use by multiple VMs. A large LUN gives VMware vSphere users the most flexibility by not requiring storage administrators to
1
http://www.vmware.com/pdf/vi3_san_design_deploy.pdf, page 93.
-9-
provision new storage every time a new VM is created. A large LUN also minimizes rescans for new LUNs on the ESX server, which can impact VMFS I/O. With its massively parallel architecture, the 3PAR InServ Storage Server uses all available disks of the same type behind its virtual volumes, which are presented as LUNs to the ESX servers. Workloads are automatically spread across all internal resources (nodes, disks, ports, etc.), delivering high application service levels and predictable levels of performance (high IOPS and low latencies). For detailed information on the 3PAR InSpire Architecture, see the 3PAR InSpire Architecture technical white paper, available from http://www.3par.com/doc_request/. With wide and autonomic distribution of workloads, storage administrators need not spend time figuring out what disks are available in an array or how best to lay out LUNs for VMware vSphere deployments. Instead, creating a virtual volume on a 3PAR InServ for the ESX Server and presenting it to the server via a LUN is a matter of a few clicks with the InForm Management Console (or a simple command with the InForm Command Line Interface) and takes less than a minute. With traditional arrays, storage administrators generally create LUNs in the 200-GB to 700-GB range for VMFS volumes, with 500-GB LUNs being a very common size. This seems to be the acceptable tradeoff between creating a large enough LUN to satisfy the performance requirements of VMware vSphere without wasting too much unused space in the LUN. However, 3PAR Thin Provisioning eliminates this tradeoff altogether.
Figure 1a. Traditional Provisioning
Figure 1b. Thin Provisioning
With 3PAR Thin Provisioning, one can allocate as much logical capacity to a VMFS volume as is needed over the lifetime of that volume without actually dedicating any physical capacity up front.
- 10 -
Physical capacity is allocated seamlessly on an as-needed basis and is consumed only when vSphere writes to the thin provisioned virtual volumes. There is no wasted capacity in the LUN that is presented to the ESX host. For this reason, compared to traditional arrays, storage utilization is very high when using Thin Provisioning. For a deeper understanding of Thin Provisioning, consult the 3PAR Thin Provisioning Benefits White Paper, available from http://www.3par.com/library.html. Aside from the storage utilization differences between a thin provisioned LUN and a traditional fat provisioned LUN, the advantage provided by the 3PAR InSpire Architecture in this situation is that both LUN types enjoy the same wide striping across all the disks in the system. Unlike many traditional approaches to thin provisioning, 3PAR Thin Provisioning does not require the creation or use of separate pools of disks within the 3PAR InServ.
4.2
Virtual Disk Format Options
When creating VMs, there are a number of options that are available for the VMDK files, not all of which are thin-friendly (and therefore compatible with 3PAR Thin Provisioning). By default, VMware vSphere creates VMs using the zeroedthick option, which is thin-friendly. With this option, when a new VM is created, the ESX Server does not immediately zero the data. It returns zeros when asked to read from unwritten areas but does not actually write the zeros to disk. For performance-intensive environments, VMware recommends using Eager Zeroed Thick virtual disks. Eager Zeroed Thick disks are also the only disk format supported with VMware Fault Tolerance (FT) as of VMware vSphere 4.0. Eager Zeroed Thick disks have the smallest overhead 2 on I/Os but require zeros to be written across all of the capacity of the VMDK at the time of creation. Thus, this VMDK format negates all the benefits of thinly provisioned LUNs since all of the physical storage is allocated when the volume is zeroed-out during creation. However, as described in the next section, using 3PAR Thin Persistence allows you to retain thin provisioning benefits when using Eager Zeroed Thick VMDKs, without sacrificing any of the performance benefits offered by this VMDK option. For a discussion on the complete list of VMware virtual disk formats, refer to 3PAR with VMware: Optimizing Virtual Environments White Paper, available from http://www.3par.com/library.html.
4.3
3PAR Thin Conversion and Thin Persistence
In a perfect world, all storage would start thin with 3PAR Thin Provisioning. However, it is never too late to become more efficient, which is why 3PAR also gives you the ability to thin existing volumes using 3PAR Thin Conversion and to keep them thin over time using 3PAR Thin Persistence software. 3PARs unique Thin Built In technology is a hardware capability built into the 3PAR Gen3 ASIC which provides an efficient, silicon-based zero-detection mechanism for converting traditional fat volumes to more efficient thin volumes on the 3PAR InServ without performance impact. This hardware capability leverages 3PARs own Thin Enginea software-based virtualization mapping engine for space reclamationwhich gives 3PAR arrays the power to remove allocated but unused space in existing volumes.
VMworld 2009 session TA2942: vSphere 4.0 Performance Best Practices.
- 11 -
Gen3 ASIC
0000 0000 0000
Before
After
Figure 2. Using 3PAR Thin Conversion Software to Thin a Volume Together with VMware Storage VMotion or any standard data migration utility, 3PAR Thin Conversion uses the zero-detection capabilities of the 3PAR Gen3 ASIC to convert "fat" volumes to thin volumes by identifying unused space (i.e., zeros) in the volume. The conversion relies on the virtualization mapping capabilities of 3PAR Thin Engine to eliminate the need to store any of the volumes unused space. Through this process, users can migrate from fat volumes on any other storage array to thin volumes on the 3PAR InServ to gain the benefits of 3PAR Thin Provisioning. For volumes that start thin (or get thin with 3PAR Thin Conversion software), the next challenge is to stay thin over time. This is where 3PAR Thin Persistence comes in. Like Thin Conversion, Thin Persistence uses Thin Built In and the 3PAR Thin Engine to reclaim unused space associated with deleted data within InServ storage volumes without disruption (Figure 3).
- 12 -
Gen3 ASIC
000
Before
After
Figure 3. Using 3PAR Thin Persistence Software to Stay Thin Over Time The process starts by activating 3PAR Thin Persistence (if necessary), then using standard file system tools (sdelete in Windows or dd in Linux) to write zeros across deleted space in a VMs file system. The zeros are autonomically detected by the 3PAR Gen3 ASIC and the disk space they were consuming is freed up and returned to the thin provisioned volume. Thus, as you delete files within VMs, Thin Persistence ensures that the volume stays thin over time. Another potential use case for Thin Persistence is when deleting a VM. Instead of deleting the VM via vCenter, you can write zeros across the VMDK file with the w option to vmkfstools. This allows the 3PAR array to reclaim the space allocated to that VM when using Thin Persistence software. Thin Persistence also ensures that, when a new, fully formatted volume is created, the entire volume is not allocated from physical storage since only zeros have been written. This situation is encountered with the Eager Zeroed Thick VMDK format since this VMDK format first zeroes out the full VMDK file and thus consumes the full size of the VMDK file on the array even before the VM has been used. With 3PAR Thin Persistence and the built-in zero-detection capability of the 3PAR Gen3 ASIC, as the ESX host writes zeros to the VMDK file, the zeros are detected in-line by the 3PAR Gen3 ASIC, and no space is allocated for the VMDK in the thin provisioned volume. Contrast this with arrays that do not offer this capability: on those arrays, an Eager Zeroed Thick VMDK negates any thin provisioning benefits because it fills up the thin provisioned volume as it is being initialized.
- 13 -
Increased Administrative Efficienc y
Virtual server environments enable increased administrative efficiency for servers, but these savings are often diluted by archaic and inflexible storage. 3PAR Utility Storage includes a number of products and features to assist the VMware administrator. 3PAR Autonomic Groups is a feature of the InForm Operating System that enables rapid provisioning to reduce storage administration time in virtualized environments by up to 90%. The 3PAR Plug-In for VMware vCenter enhances visibility into storage resources for the VMware administrator by integrating 3PAR-specific information into vCenter.
5.1
3PAR Plug-In for VMware vCenter
The 3PAR Plug-In for VMware vCenter is a vSphere management console plug-in that allows easy identification of 3PAR virtual volumes used by VMs and datastores. It provides an integrated view of the VMs and associated storage resources. Properties such as volume type (Fat or TPVV), device type (FC disk, NL disk, or SSD), RAID level, etc. are displayed via the 3PAR tab in the vSphere management console. The 3PAR Plug-In for VMware vCenter provides the VMware administrator a view into the InServ via a single pane (Figure 4). There is no need to log in to the InServ to identify space consumption or determine how a volume maps to a datastore. This information is easily visible via the 3PAR Plug-In, as is capacity usage and other data.
- 14 -
Figure 4. Viewing Virtual Volume Mapping Information via the 3PAR Plug-In
5.2
3PAR Recovery Manager for VMware vSphere
3PAR Recovery Manager for VMware vSphere enables the protection and restoration of VMs and datastores (Figure 5). It provides virtual copy management and allows the administrator to take LUN-level snapshots of VMs and datastores via the vSphere management console. 3PAR Recovery Manager provides array-based snapshots that are fast, space-efficient, and VM-aware. This 3PAR plug-in solves the issues associated with traditional, agent-based backup schemes that are typically slow, complex, and fail to offer the flexibility and granularity that array-based snapshots can provide. 3PAR Recovery Manager makes it possible to create hundreds of virtual copies. The number of virtual copies to retain and the retention period for each virtual copy can easily be specified. Once a virtual copy has been created, this 3PAR plug-in allows the flexibility of granular restores at the VMFS layer, the VM layer, or the individual file level. For more detailed information on the 3PAR Plug-In for VMware vCenter or 3PAR Recovery Manager for VMware vSphere, see http://www.3par.com/solutions/virtualization.html.
- 15 -
Figure 5. Virtual Copy of a VM as Displayed from Inside the vSphere Management Console
5.3
3PAR Replication Adapter for VMware vCenter Site Recovery Manager (SRM)
VMware vCenter Site Recovery Manager (SRM) provides end-to-end management of arraybased replication, virtual machine failover, and automated disaster recovery management for environments that use VMware vCenter Server. The 3PAR Replication Adapter for VMware vCenter SRM was developed to provide integration between SRM and 3PAR Remote Copy. This integrated solution is detailed in Implementing 3PAR Remote Copy with VMware Site Recovery Manager, available from http://www.3par.com/library.html.
5.4
3PAR Dynamic Optimization and VMware vSphere
3PAR Dynamic Optimization is an optional software feature that offers an online and nondisruptive way to make changes to volumes on the 3PAR array. Storage administrators can move volumes between different disk types (Fibre Channel, Nearline), convert between RAID levels (RAID 1, RAID 5, or RAID 6), and/or rebalance volumes as new disks are added, all without impacting any hosts that the InServ is busy serving. In a VMware environment, 3PAR Dynamic Optimization can be used to move running VMs from Fibre Channel disks to Nearline disks without impacting what the VMs are doing. Similarly, as new disks are added to the array, the LUN that ESX is using can be striped across the new disks on the fly without taking an outage at the ESX server level.
- 16 -
VMware VMotion offers somewhat similar functionality, but at the VMware layer. 3PAR Dynamic Optimization works at the storage layer, bypassing any application-level overheads. It can be used to optimize data service levels while VMware VMotion can be used to optimized CPU processing levels.
5.5
Boot-from-SAN Environments
In a boot-from-SAN environment (Figure 6), each ESX servers operating system is installed on a volume on the array, not on the servers internal disk. In this situation, you should create a separate virtual volume for each ESX server to be used for the boot image. The boot from SAN approach brings a number of benefits to a VMware vSphere environment: 3PAR Virtual Copy can be used to take snapshots of the boot image and provide backout capabilities when upgrading ESX. Users can toggle between different ESX versions by simply pointing the ESX servers HBA to the different LUNs containing the respective ESX versions. Replacing an ESX server simply means pointing the new server to the boot LUN. Tech Tip: The steps to set up an ESX server to boot from SAN at a high level are as follows: 1. Boot to the HBA BIOS 2. Enable the HBA port to boot 3. Perform zoning 4. Create and present the VV from the InServ 5. Discover the LUN and mark it as bootable in the HBA BIOS Note: VMwares Fibre Channel SAN Configuration Guide details how to accomplish this for QLogic and Emulex HBAs.
Figure 6. A Boot-from-SAN Configuration
- 17 -
Conclusion
Deploying 3PAR Utility Storage in VMware vSphere environments helps remove the management, provisioning, and monitoring headaches associated with traditional storage arrays. Traditional SAN storage struggles to meet the performance and agility needs of a virtualized server environment, whereas 3PAR Utility Storage solves the heavy demands that server virtualization places on storage by using massive parallelization for exceptional performance and high availability features for superior resilience that enables you to consolidate with confidence. Support for the latest version of VMware vSphere and the 3PAR Plug-In for VMware VAAI deliver enhanced performance, agility, and scalability in vSphere environments while 3PAR Thin Provisioning allows physical storage to be consumed only when required for actual written data, rather than when allocated. 3PAR Dynamic Optimization can be used to tailor storage performance without disruption to VMware vSphere. These are among the unique advantages that make 3PAR Utility Storage the perfect foundation for building or expanding a virtualized server environment with VMware vSphere.
- 18 -
Appendix: Frequentl y Asked Questions

The following questions commonly arise when integrating 3PAR InServ Storage Servers with VMware vSphere: Q: Should thin provisioned virtual volumes (TPVVs) be used for ESX datastores? A: Thin Provisioning makes sense as long as you are not going to fill up the volume right away. If the volume is going to be full in a short time, there is no benefit to be gained from thin provisioning. If VMware snapshots are being created and are not being cleaned up, the benefits of thin provisioning will be negated. To the ESX host, it makes no difference if the host sees a thin provisioned LUN or a traditional fat provisioned LUN. Q: Is there any overhead to using TPVVs? A: The additional overhead of TPVVs as compared to traditional volumes is negligible. Q: If using a 2-TB TPVV, what will VMware see? A: VMware will see a 2-TB LUN available for its use. It doesnt know the difference between a thin provisioned virtual volume and a traditional virtual volume. Q: What size VVs should be created? A: The volume size is not as important to 3PAR, as VVs are widely striped across as many drives as possible within the InServ array. If using 3PAR Thin Provisioning, actual storage capacity is only consumed upon write. However, a 2-TB VV will be able to accommodate many more VMs than a 500-GB VV. For ESX 4.1 or later, create a VV that is as large as you are comfortable with since vSphere 4.1 has improved the way it performs metadata locking (see section 3.4). For ESX 4.0 and prior, in order to minimize the impact of SCSI reservations (see section 3.3) and keep the environment well balanced, it is best to create 500-GB volumes (thin provisioned or fat provisioned). Q: How many VMs can be put on a single VV? A: It depends! There is no one answer that will work for every situation. A number of factors such as server hardware, number of CPUs, amount of memory, type of VMs, applications running in the VMs, etc. will determine how many VMs can be hosted on a LUN comfortably. Q: What type of path policy (Fixed, MRU, Round Robin) should be used with 3PAR InServ Storage Servers? A: With ESX 3.5, use the default policy (Fixed). With ESX 4.0, change the default policy to the Round Robin path policy. To do this, log in to the service console for each ESX 4.0 host and run the following command line:
esxcli nmp satp setdefaultpsp --satp VMW_SATP_DEFAULT_AA --psp VMW_PSP_RR
Q: In ESX 4.0, VMware now supports thin VMs. Whose thin provisioning should be used: VMware, 3PAR, or both? A: VMware thin provisioning only applies to VMs at the VMFS level. It allows one to overallocate VMs to maximize VMFS usage. If the goal is to reduce storage costs and maximize storage utilization, then use 3PAR Thin Provisioning to provision large VMFS volumes with minimal upfront storage costs. There are no additional storage savings to be realized by using VMware thin provisioning. VMware thin provisioning does consume some CPU cycles on the ESX host as it is performed at the software layer (as compared to 3PAR Thin Provisioning, which is performed on the array). It is perfectly fine to place VMware thin VMs on 3PAR Thin Provisioning volumes with
- 19 -
the only caveat of having to manage thin provisioning at both the VMware level as well as the array level.
- 20 -
About 3PAR
3PAR (NYSE: PAR) is the leading global provider of utility storage, a category of highly virtualized, dynamically tiered, multi-tenant storage arrays built for public and private cloud computing. Our virtualized storage platform was built from the ground up to be agile and efficient and to eliminate the limitations of traditional storage arrays for utility infrastructures. As a pioneer of thin provisioning and other storage virtualization technologies, we design our products to reduce power consumption to help companies meet their green computing initiatives and to cut storage total cost of ownership. 3PAR customers have used our self-managing, efficient, and adaptable utility storage systems to reduce administration time and provisioning complexity, to improve server and storage utilization, and to scale and adapt flexibly in response to continuous growth and changing business needs. For more information, visit the 3PAR Website at: www.3PAR.com.

VMW WP 10.1

Загружено:

Сведения о документе

Исходное описание:

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

VMW WP 10.1

Загружено:

Авторское право:

Доступные форматы

VMware vSphere with 3PAR Utility Storage

Appendix: Frequently Asked Questions ...................................................................................... 18

Performance and Reliability Features

Queue Depth Throttling

Metadata Locking with SCSI Reservations

VMware vStorage APIs for Array Integration (VAAI)

Storage I/O Controls

VMFS versus RDM

Virtual Volumes: Thin or Fat?

http://www.vmware.com/pdf/vi3_san_design_deploy.pdf, page 93.

Figure 1a. Traditional Provisioning

Figure 1b. Thin Provisioning

Virtual Disk Format Options

3PAR Thin Conversion and Thin Persistence

VMworld 2009 session TA2942: vSphere 4.0 Performance Best Practices.

Increased Administrative Efficienc y

3PAR Plug-In for VMware vCenter

3PAR Recovery Manager for VMware vSphere

3PAR Dynamic Optimization and VMware vSphere

Figure 6. A Boot-from-SAN Configuration

Appendix: Frequentl y Asked Questions

Вам также может понравиться