You are on page 1of 49

SQL Server 2000 Analysis Services

Patrick Brulotte Consultant Microsoft Consulting Services

Agenda

Analysis Paralysis OLAP to the Rescue OLAP and Data Warehousing Analysis Services Architecture Analysis Services Applied

Sales Analysis Financial Modeling Internet Analytics

Analysis Paralysis
Sound Familiar?

I Have Too Much Data and Not Enough Information Why Dont All of My Reports Tie Together? Management Needs Answers Yesterday Who Has Time for Analysis?

Analysis Paralysis
You Are Not Alone

Available Business Data is Growing Exponentially Critical Decisions Per Week Have Tripled Over the Last Five Years Dedicated Analytical Personnel Have Decreased by 50% Over the Last 5 Years
Source: Gartner Group

Analysis Paralysis

The Problem: Systems Arent Integrated

Analysis Paralysis
The Result: Pain and Suffering

Erroneous Assumptions

Different definitions of a measure by two systems Incomplete information Time consuming Error prone Management stops asking the tough questions

Re-Keying of Data

Not Enough Time for Analysis

Agenda

Analysis Paralysis OLAP to the Rescue OLAP and Data Warehousing Analysis Services Architecture Analysis Services Applied

Sales Analysis Financial Modeling Internet Analytics

OLAP to the Rescue


OLAP Bridges the Analysis Gap

OLAP
Relational Databases
Store large amounts of data (OLTP) Require IT knowledge Contain minimal computational/ analysis capability

Data consistency Multi-dimensional Large data volumes Quick query times Powerful analytics Strong calculations End-user friendly

Spreadsheets
Analyze small amounts of data Distributed among front-line end users No corporate data consistency

OLAP to the Rescue


OLAP Definition

On-Line Analytical Processing


Versus OLTP Term coined by E.F. Codd

Generally Synonymous with EIS, DSS, Business Intelligence Multi-Dimensional View of Data Intuitive Data Model Performance is Key

OLAP: A Business Decision View of Data


Location

Atlanta Denver Detroit


1,930 UNITS $6,745 GROSS $5,831 COST

Grapes

Product

Cherries
Melons Apples Pears

Q1

Q2 Q3
Time

Q4

Agenda

Analysis Paralysis OLAP to the Rescue OLAP and Data Warehousing Analysis Services Architecture Analysis Services Applied

Sales Analysis Financial Modeling Internet Analytics

OLAP and Data Warehousing


The Big Picture
Data Warehouse
OLAP

OLTP Source

Transformation
Data Mart

External Data

OLAP Is the User Deliverable

End Users

OLAP and Data Warehousing


The Relational Component

Fact Table Dimension Table Star Schema Snowflake Schema

OLAP and Data Warehousing


Fact Table
Date Product Customer Units Dollars

Dimensions

Measures

The Fact Table Is the Relational Version of a Cube

OLAP and Data Warehousing


Fact Table
Date 1/2/99 1/2/99 1/2/99 1/2/99 1/3/99 1/3/99 1/3/99 1/3/99 SKU 101 101 102 102 101 101 102 102 Cust Jones Smith Jones Smith Jones Smith Jones Smith Units 250 92 105 81 125 302 144 171 Dollars $3,295 $1,422 $1,750 $1,090 $2,105 $3,988 $2,675 $3,009

OLAP and Data Warehousing


Dimension Table
Region_ID Region_Name Territory 101 102 103 Northwest Northeast Southwest North North South Manager Jane Sam Paul

104

Southeast

South

Amy

Dimension Tables Provide Descriptions, Hierarchies, and Member Properties

OLAP and Data Warehousing


A Star Schema

A Star Schema Consists of

A Central Fact Table

Joined to related Forming the shape of a

Dimension Tables

Star

OLAP and Data Warehousing


Star Schema

Fact Table

Dimension Table

OLAP and Data Warehousing


Using DTS

OLTP Source

DTS
External Source

DW Storage

OLAP OLE DB Client Server for OLAP, Apps ADO-MD

Agenda

Analysis Paralysis OLAP to the Rescue OLAP and Data Warehousing Analysis Services Architecture Analysis Services Applied

Sales Analysis Financial Modeling Internet Analytics

Analysis Services Architecture

DTS OLTP Source DW Storage

DW Mgr

OLAP Server

C/S Interface

Client Apps

Analysis Services Architecture


Architecture
Analysis Manager
SQL Server Data Warehouse

Analysis Services

OLEDB for OLAP OLE DB for DM

PivotTable Service

DSO

Application

Other OLE DB Providers

Analysis Server
OLAP Store
OLAP Engine Data Mining Engine

TCP/IP HTTP

ADO MD

Processing

Querying

Analysis Services Architecture


MS OLAP Flexible Storage

Users and Applications See Only Cubes

Analysis Services Architecture


Partial Aggregation
Show all sales for all products for all . . .
Highest level of aggregation

Most detailed aggregations Fact Table

Agenda

Analysis Paralysis OLAP to the Rescue OLAP and Data Warehousing Analysis Services Architecture Analysis Services Applied

Sales Analysis Financial Modeling Internet Analytics

Sales Analysis
Graphic Design Institute - Sales

E-Commerce Everything Graphics Dimensions


Time (down to the day) Geography (Ship-to) Customer Products

Sales Analysis
Distinct Count

High End Feature Difficult in OLAP Systems

Customer Product Day jsmith Wine 4/16/99

dharris jsmith
kyoung dharris dharris jsmith dharris

Beer Wine
Wine Wine Wine Beer Beer

4/16/99 4/16/99
4/16/99 4/17/99 4/17/99 4/17/99 4/17/99

How many DISTINCT customers bought wine and beer each day?
All Products Wine All Time 4/16/99 4/17/99 6 3 3 3 2 3 2 1 Beer 3 2 1 2

Sales Analysis
Drillthrough
Store Sales January, 1997 Customer Beer Wine Bread Jones 66 300 20 Smith 10 5 White 100 10 25 Date Customer

1-Jan Jones 5-Jan Jones 9-Jan Jones 10-Jan Jones 17-Jan Jones 24-Jan Jones

Product Amount Heineken 6 Pack 6.5 Case Coors 20 Samuel Adams 6 7.25 Samuel Adams 6 6.75 Miller Light 6 Pack 5.75 Case Coors 20

Sales Analysis
Summary

Distinct Count Measure Drill Through Architecture Wizard-Driven Interface

Agenda

Analysis Paralysis OLAP to the Rescue OLAP and Data Warehousing Analysis Services Architecture Analysis Services Applied

Sales Analysis Financial Modeling Internet Analytics

Financial Modeling
Regular Dimensions
USA

Country
WA

CA

ID

State
Bothell

SF

LA

Boise

Seattle

Redmond

City

Uniform Levels

Depth All members at a given level have the same number of ancestors

No Gaps

Financial Modeling
Ragged Dimensions
All

USA

Israel

Country State City

CA

WA

No States!
Tel Aviv Haifa

SF

LA

Redmond

Non-Constant Depth States/Provinces Are a Good Example

Financial Modeling
Smith

Parent-Child Dimensions
Employee Smith Jones White Block Hart Knight Fox Hunt Smart Manager <none> Smith Smith Jones Jones Jones Hart Hart Hunt

Jones

White

Block

Hart

Knight

Fox

Hunt

Smart

Financial Modeling
Custom Rollups and Formulas
+ Net Income + Net Sales + Revenue - Discounts - Total Expenses + Direct + Indirect ~ Allocated

Financial Modeling
Multi-Dimensional Expressions (MDX)

Query Language Calculation Syntax Security Actions Part of OLE DB for OLAP Spec

TM1, SAS, Analysis Services, Whitelight, etc.

Exploits Multi-Dimensional Structure The KEY for Advanced Analysis

Financial Modeling
Write-Back

Cube Must Be Enabled for Write-Back Writes at Lowest Level, or Summary Level Does Not Write Directly to Cube or Fact Table Writes to (and Reads from) an Incremental Change Table Virtually No Recalc Time

Financial Modeling
Security

Cell-Level Security Dimension Member Security

Cell-Level Security
Employees Total salary East 50 N/A Sq Ft 16,500

Member-Level Security
Employees Sq Ft 50 75 35 16,500 20,625 10,500

Central
West

75
35

$7,600
N/A

20,625
10,500 East Central West

Financial Modeling
Summary

Ragged Dimensions Parent-Child Dimensions Custom Rollups and Formulas Write-Back at Bottom Level and Rollup Levels MDX Security

Agenda

Analysis Paralysis MS OLAP to the Rescue OLAP and Data Warehousing Analysis Services Architecture MS OLAP Applied

Sales Analysis Financial Modeling Internet Analytics

Internet Analytics
Key Challenges

Very Large Data Volumes Very Large Dimensions No Natural Hierarchies for Some Dimensions External Providers Collect Clickstream Information

How to get access to this data over the internet

Integrate Click-Stream Data with Operational Data with Non-Numeric Data

Internet Analytics
Partitioning to Scale
Current Year
MOLAP 35% Agg HOLAP 20% Agg

Prior Year History

Multiple Aggregation Designs Multi-Server Support

ROLAP 0% Agg

Internet Analytics
Very Large Dimensions Scalability

Large MOLAP Dimensions

Non-memory resident (up to 10 million members) 100s of millions of members

Huge ROLAP Dimensions

Low Level Support for Virtual Dimensions


No impact on data storage. No scalability or dimension cardinality issues Multiple hierarchy levels

Internet Analytics
Virtual and Linked Cubes

Combine Local and Remote Cubes Internal and External Cubes

East

Sales
West

3rd-Party External

Internet Analytics
Data Mining

Advanced Analytical Capabilities


Clustering Algorithm Decision Tree Algorithm Plug-in your own

Have the System Discover Patterns in the Data Uses OLAP or Relational Data

Internet Analytics
Summary

Partitioning ROLAP Dimensions Virtual Cubes Linked Cubes Access via the Internet Data Mining

Call to Action

Build OLAP Solutions Using SQL Server 2000

Stop the analysis paralysis Enable business intelligence in your company Take advantage of centralized business rules and calculations Utilize the robust multi-dimensional engine

Recommended Resources from Microsoft Press

Microsoft SQL Server 2000 Programming Step-by-Step

Pick up your CD & Book at todays event and get a portion of this and other Microsoft resources for developers
Microsoft Press Resources available through resellers and on the Web at http://mspress.microsoft.com/developer

More Resources
www.microsoft.com/sql/productinfo/olap.htm www.olaptrain.com www.olapreport.com www.microsoft.com/sql/ msdn.microsoft.com www.microsoft.com/sql/bizsol/datawarehousing.htm

You might also like