Oracle OLAP

Ratan Vakil
Business Analytics, OLAP
Ratan.Vakil@oracle.com
Aim or yahoo chat: ofaguru
 The Business Requirements
Who generates them?

Competitive Analysis- Just about

Key Indicator Tracking-
Trend Analysis- everyone in the
Exception Reporting-
Board company!!

Investment/Acquisitions- -Resource Allocation

Reorganization Analysis- Planning / -Capacity Planning
Long Range Planning- -Human Resource Planning
Analysts

Budgeting- Sales & -Sales Tracking

Consolidation- Finance Marketing -Promotion Analysis
Variance Analysis- Dept -Sales Forecasting
Dept

Financial Modelling- -Product Profitability

Foreign Exchange/Cash Management- -Customer Profiling
Asset Liability Modelling- -Distribution Analysis
Activity Based Management- -Sales Performance/Effectiveness
 OLTP vs. OLAP
Transactional Analytical

Who? Why?
What? How?
Where? What if ?

Report data Analyze & use data

Pedigree (short) …
OLAP is alive and well at Oracle

y Over thirty years of innovation yields a

complete and compelling OLAP platform
– Express, the first multidimensional database
– Oracle 9iR2, the first (and only) relational-
multidimensional database
– Oracle 10g
y The first (and only) Grid capable OLAP platform
y All new administration
y All new data access tools
y All new applications
 Purely Relational, ROLAP, or MOLAP?
A typical MOLAP implementation

What if …
– A single database offered the openess of a relational
solution?

– … and provided the calculation power of a

multidimensional engine ?

– The calculations could be defined as easily as spreadsheet

formulas?

– The system was efficient to build and maintain ?

– Users experienced excellent query performance ?

 The Business Requirements
Why do they need OLAP?

Businesses need OLAP because:

• Multidimensional model: They want to inspect their data in a
multidimensional format that includes dimensions, levels, hierarchies
and attributes.
• Calculations: They want to define calculations that adhere to the
proprietary rules that govern their particular multidimensional view of
the data. For example, as in aggregation.
• Processing efficiency: Since analysis is an intensively re-iterative
process, the query response time must be sub-second. OLAP
engines are better designed to meet this requirement.
• Transaction model: A read-repeatable transaction model that
supports what-if analysis.
Design – Logical models
coc_Officer_Dim coc_Measure_Result_Graph
PK ASSOC_ID Lookup PK,FK1 ASSOC_ID
PK ROLE_CD PK MEASURE_ID

coc_Customer_Dim coc_Product_Dim

PK CST_ID coc_Profitability PK PRD_CD

PK,FK1 CST_ID
PK,FK2 ASSOC_ID
PK,FK2 ROLE_CD
PK,FK3 PRD_CD
PK,FK4 SYS_ASOF_DT coc_Time_Dim
coc_Officer_Dim PK,FK4 PERIOD_CD
PK SYS_ASOF_DT
PK,FK4 SOURCE_DT
PK ASSOC_ID PK PERIOD_CD
PK ROLE_CD PK SOURCE_DT

coc_Customer_Dim
PK CST_ID coc_Opportunity
PK,FK1 CST_ID
PK,FK2 ASSOC_ID
PK,FK2 ROLE_CD
PK,FK3 SYS_ASOF_DT coc_Time_Dim
PK,FK3 PERIOD_CD
coc_Officer_Dim
PK,FK3 SOURCE_DT PK SYS_ASOF_DT
PK ASSOC_ID PK PERIOD_CD
PK ROLE_CD PK SOURCE_DT

8
 Purely Relational, ROLAP, or MOLAP?

Select a purely relational implementation when …

• The analytic requirements of the business are

met by the capabilities of SQL.

• There are appropriate in-house SQL skills.

• The relational engine provides satisfactory

query performance.
 Purely Relational, ROLAP, or MOLAP?
Relational Technology

PRODUCT CUST
• prod A • cust A
• prod B • cust B
• prod C • cust C
• prod D • cust D
• prod E • cust E
SALES
• Prod
• Cust
• Time

TIME
• Year
• Quarter
• Month
• Week
• Day

A purely relational implementation is designed and optimized to support the

efficient movement and calculation of large volumes of data.
 Purely Relational, ROLAP, or MOLAP?

Select a ROLAP implementation when …

• The analytic requirements of the business are

met by the capabilities of SQL.

• User is looking for an easier way to formulate

complex queries.

• The detail data is very sparse.

· Use Materialized Views to optimize performance.

 Purely Relational, ROLAP, or MOLAP?
Relational Technology
Dimension Cube Dimension
Object
Object

PRODUCT CUST
• prod A • cust A
• prod B • cust B
• prod C • cust C
• prod D • cust D
• prod E • cust E
SALES
• Prod
• Cust
Dimension • Time
Object
TIME Dimension Object
• Year • Dimension
• Quarter • Hierarchy
• Month
• Week • Level
• Day • Attribute

This is a ROLAP IMPLEMENTATION.

 Purely Relational, ROLAP, or MOLAP?

Select a MOLAP implementation …

• When the analytic requirements of the business need the

extended analytic, forecasting and planning functionality of
Multidimensional Database Technology.

• When the analysis includes lots of calculated and aggregated

Key Performance Indicators

• Need an easier way to define complex or proprietary

calculations.

• Need a transaction model that supports what-if analysis.

Purely Relational, ROLAP, or MOLAP?
Multidimensional Technology

Cube

$ $ $ $
SALES dimensioned by
$ $ $ $ PRODUCT,
$ $ $ $ CUSTOMER,
TIME
$ $ $ $

This is a MOLAP implementation

 Purely Relational, ROLAP, or MOLAP?
Multidimensional Technology

Some benefits of the multidimensional processing model . . .

• A separate query is formulated and executed for each

dimensional component of the query.
‘Ease of use feature!

• No JOIN is required when using the multidimensional

technology.
‘Improved performance!

• “Aggregate then filter” methodology is used.

‘Consistent, correct results.
‘Intelligent drill!
‘.
 Architecture
The OLAP Option in the Oracle Database

Components include:

• A powerful SQL calculation engine

• A powerful multidimensional calculation engine
• Multidimensional data storage and retrieval
• Programming APIs for SQL, PL/SQL, and Java
• Dimensionally aware data manipulation language
(OLAP DML)
• SQL access to multidimensional data
 Prior Architecture

DB2, SQL
Server,
Oracle, etc.
Clients

Operational Data Warehouse Data Marts Multidimensional

Sources Databases

Warehouse ETL

OLAP ETL

Runtime
 Current Architecture
The OLAP Option in the Oracle Database

Oracle Database
Oracle Call Interface JDBC

Metadata OLAP API

Relational Technology Object Technology OLAP Technology

Multidimensional
SQL Engine Table Functions
Engine

Storage

Relational Data
Multidimensional
Data
 Architecture – Open access

Because no single tool will satisfy all of

the users in an organization …

… Oracle has products that represent

every class of reporting tool.
 Architecture – Open access

8. Any SQL
Any SQL
interface or query HTML DB
Query Tool tool
7. Custom built 1. Simple reporting tools
SQL, PL/SQL
applications that execute pre-defined
OLAP Java API SQL statements, and has no
knowledge of OLAP
Forms Builder.lnk
HTML DB
Oracle Forms

6. Spreadsheet tools DW 2. OLAP-aware tools

Oracle Excel ETL
that generate SQL
Add-in OLAP
Discoverer Plus
OLAP Excel
Data Mining
Add-in Discoverer
Plus
5. OLAP-aware 3. Multidimensional
application object-aware
building tools OLAP-aware tools that
BI Beans generate SQL
4. Tools that provide
Discoverer
the ability to define
Plus OLAP
highly formatted reports
Oracle Reports in multiple formats
(OLTP) Discoverer
BI Beans Plus OLAP
Reports
 Architecture – Open access

8. Any SQL
Any SQL
interface or query HTML DB
Query Tool tool
7. Custom built 1. Simple reporting tools
SQL, PL/SQL
applications that execute pre-defined
OLAP Java API SQL statements, and has no
knowledge of OLAP
Forms Builder.lnk
HTML DB
Oracle Forms
Every Oracle
6. Spreadsheet tools tool can access
DW 2. OLAP-aware tools
Oracle Excel the power of the
ETL
that generate SQL
Add-in AnalyticOLAP
Discoverer Plus
OLAP Excel Workspace.
Data Mining
Add-in Discoverer
Plus
5. OLAP-aware 3. Multidimensional
application object-aware
building tools OLAP-aware tools that
BI Beans generate SQL
4. Tools that provide
Discoverer
the ability to define
Plus OLAP
highly formatted reports
Oracle Reports in multiple formats
(OLTP) Discoverer
BI Beans Plus OLAP
Reports
 Architecture – Open access

8. Any SQL
Any SQL
interface or query HTML DB
Query Tool tool
7. Custom built 1. Simple reporting tools
SQL, PL/SQL
applications that execute pre-defined
OLAP Java API SQL statements, and has no
knowledge of OLAP
Forms Builder.lnk
In fact, the HTML DB
Oracle Forms power of the
6. Spreadsheet tools Analytic DW 2. OLAP-aware tools
Oracle Excel Workspace can that generate SQL
ETL
Add-in be accessed by
OLAP
Discoverer Plus
OLAP Excel any third-party
Data Mining
Add-in tool that emits Discoverer
Plus
5. OLAP-aware SQL! 3. Multidimensional
application object-aware
building tools OLAP-aware tools that
BI Beans generate SQL
4. Tools that provide
Discoverer
the ability to define
Plus OLAP
highly formatted reports
Oracle Reports in multiple formats
(OLTP) Discoverer
BI Beans Plus OLAP
Reports
 A Closer Look

Application

OLAP API

This object Metadata

represents the
Analytic SQL Generator
Workspace
select … from view or table

Let’s take a
closer look at Row and Table ADTs
the AW.
Relational View(s)

OLAP_TABLE

Relational Multidimensional
 A Closer Look
Analytic Workspace

• An Analytic Workspace is a container that

holds multidimensional data and objects.

• The data in the AW is manipulated by the

multidimensional calculation engine that is
imbedded in the RDBMS.

• AWs and the multidimensional engine were

designed for efficient processing of
multidimensional calculations.
 A Closer Look
Analytic Workspace

While A
Do B
End
Program
source code Dimension definitions
Measure definitions
OLAP
DML

Hierarchy Data
definitions Formulas and
relationships
equations

Dimension Measure
Definitions for logical
data data
groupings of data
-- CUBES
CUBES --
 Prepare the OLAP Data Warehouse
The tools

Use these tools to design and build your OLAP

data warehouse:
• Oracle Warehouse Builder – End-to-end ETL tool
• Enterprise Manager – Describe the star or snowflake logical data
model

• Analytic Workspace Manager – Build AW from star schema

• DBMS_AWM APIs – Build AW from star schema
• OLAP DML – Programmatically build the AW and all of its objects
Dimensional Model
Dimensional Model
Q U E S T I O N S
A N S W E R S