aggregating knowledge in a data warehouse and...
TRANSCRIPT
Aggregating Knowledge in a Data Warehouse and Multidimensional AnalysisRafal LukawieckiStrategic Consultant, Project Botticelli Ltd
2
Objectives
• Explain the basics of:
1. Data Warehousing
2. ETL
3. OLAP/Multidimensional Data
• Relate the theory to SQL Server 2008 SSAS and SSIS
This seminar is based on a number of sources including a few dozen of Microsoft-owned presentations, used with permission. Thank you to Marin Bezic, Kathy Sabourin, Aydin Gencler, Bryan Bredehoeft, and Chris Dial for all the support. Thank you to Maciej Pilecki for assistance with demos.
The information herein is for informational purposes only and represents the opinions and views of Project Botticelli and/or RafalLukawiecki. The material presented is not certain and may vary based on several factors. Microsoft makes no warranties, express,implied or statutory, as to the information in this presentation.
Portions © 2009 Project Botticelli Ltd & entire material © 2009 Microsoft Corp. Some slides contain quotations from copyrightedmaterials by other authors, as individually attributed or as already covered by Microsoft Copyright ownerships. All rights reserved.Microsoft, Windows, Windows Vista and other product names are or may be registered trademarks and/or trademarks in the U.S.and/or other countries. The information herein is for informational purposes only and represents the current view of Project BotticelliLtd as of the date of this presentation. Because Project Botticelli & Microsoft must respond to changing market conditions, it shouldnot be interpreted to be a commitment on the part of Microsoft, and Microsoft and Project Botticelli cannot guarantee the accuracy ofany information provided after the date of this presentation. Project Botticelli makes no warranties, express, implied or statutory, as tothe information in this presentation. E&OE.
3
1. Data Warehouse
4
OLE DB
ODBC
DB2Oracle
XML
SQL Server
Analysis Services
SQL Server
Report Server Models
SQL Server
Data Mining Models
SQL Server
Integration Services
MySAP
Hyperion Essbase
SAP
NetWeaver BISQL Server
Teradata
Rich ConnectivityData Providers
5
Let’s Store the Intelligence: DW
• SQL Server Analysis Services server is a logical endpoint for data being aggregated with SSIS
• But do not store actual data in it
• Data physically rests in another database called a Data Warehouse
• You can manipulate it directly, or build in parallel with OLTP processing
• Modelling of data stored in DW and analysed using SSAS is at the heart of good Data Warehouse design
6
Star Schema
7
Star Schema Benefits
• Transforms normalized data into a simpler model
• Delivers high-performance queries
• Delivers higher performing queries using Star Join Query Optimization
• Uses mature modeling techniques that are widely supported by many BI tools
• Requires low maintenance as the data warehouse design evolves
8
Snowflake Dimension Tables
• Define hierarchies using multiple dimension tables
• Support fact tables with varying granularity
• Simplify consolidation of data from multiple sources
Potential for slower query performance in relational reporting
No difference in performance in Analysis Services database
9
Hierarchies
• Benefits
• View of data at different levels of summarization
• Path to drill down or drill up
• Implementation
• Denormalized starschema dimension
• Normalized snowflakedimension
• Self-referencing relationship
10
Fact Table Fundamentals
• Collection of measurements associated with a specific business process
• Specific column types
• Foreign keys to dimensions
• Measures – numeric and additive
• Metadata and lineage
• Consistent granularity – the most atomic level by which the facts can be defined
11
Fact Table Examples
Day Grain
Quarter Grain
Reseller sales data by:
•Product
•Order Date
•Reseller
•Employee
•Sales Territory
Sales quota data by:
•Employee
•Time
12
Date Dimension Table
• Most common dimension used in analysis (aka Time dimension)
• Used consistently with all facts for efficient and flexible analysis
• Useful common attributes – Year, Quarter, Month, Day
• Time series analysis support
• Navigation and summarization enabled with hierarchies, such as calendar or fiscal
• Single table design (typically not snowflake design)
Tip: Format the key of the dimension as yyyymmdd (e.g. 20060925) to make it readily understandable
13
Parent-Child Hierarchy
• A dimension that contains a parent attribute
• A parent attribute describes a self-referencing relationship, or a self-join, within a dimension table
• Common examples
• Organizational charts
• General Ledger structures
• Bill of Materials
14
Parent-Child Hierarchy Example
Brian
Amy
Stacia
Stephen
Shu Michael
Peter
José
Syed
15
Slowly Changing Dimensions
• Support primary role of data warehouse to describe the past accurately
• Maintain historical context as new or changed data is loaded into dimension tables
• Implement changes by Slowly Changing Dimension (SCD) type
• Type 1: Overwrite the existing dimension record
• Type 2: Insert a new ‘versioned’ dimension record
• Type 3: Track limited history with attributes
16
SCD Type 1
• Existing record is updated
• History is not preserved
17
SCD Type 2
• Existing record is ‘expired’ and new record inserted
• History is preserved
• Most common form of SCD
18
SCD Type 3
• Existing record is updated
• Limited history is preserved
• Implementation is rare
18
SalesTerritoryKeyupdate to 10
19
Let’s Get the Data
• We would like to populate facts and dimensions in our Data Warehouse from OLTP data...
20
2. Integration and ETL
21
Let’s do ETL with SSIS
• SQL Server Integration Services (SSIS) service
• SSIS object model
• Two distinct runtime engines:
• Control flow
• Data flow
• 32-bit and 64-bit editions
22
The Package
• The basic unit of work, deployment, and execution
• An organized collection of:
• Connection managers
• Control flow components
• Data flow components
• Variables
• Event handlers
• Configurations
• Can be designed graphically or built programmatically
• Saved in XML format to the file system or SQL Server
23
Control Flow
• Control flow is a process-oriented workflow engine
• A package contains a single control flow
• Control flow elements
• Containers
• Tasks
• Precedence constraints
• Variables
24
Data Flow
• The Data Flow Task
• Encapsulates the data flow engine
• Exists in the context of an overall control flow
• Performs traditional ETL in addition to other extended scenarios
• Is fast and scalable
• Data Flow Components
• Extract data from Sources
• Load data into Destinations
• Modify data with Transformations
• Service Paths
• Connect data flow components
• Create the pipeline
25
Data Flow Sources
• Sources extract data from
• Relational tables and views
• Files
• Analysis Services databases
26
Data Flow Destinations
• Destinations load data to
• Relational tables and views
• Files
• Analysis Services databases and objects
• DataReaders and Recordsets
Enterprise Edition only
27
Populating Fact Tables
Y
Insert newrecord
Insert newdimension record
Lookupdimension key
N
Lookup failed?
Repeat for each
dimension key
TransformFact
source
28
Populating Dimension Tables
Y
Insert newrecord
Update changedcolumn(s)
Expire existingrecord
TransformCorrelaterecords
N
N
Y
Type 2change?
Y
Type 1change?
New record?
Dimension
source
29
Data Flow Transformations
• Aggregate, merge, distribute, or modify data
• Include error outputs in some cases
Transformation Categories
Row
Rowset
Split and Join
Business Intelligence (BI)
Script
Other
30
Row Transformations
• Update column values or create new columns
• Transform each row in the pipeline input
31
Rowset Transformations
• Create new rowsets that can include
• Aggregated values
• Sorted values
• Sample rowsets
• Pivoted or unpivoted rowsets
• This is a heavy-weight performer of SSIS
• Are also called asynchronous components
32
Split and Join Transformations
• Distribute rows to different outputs
• Create copies of the transformation inputs
• Join multiple inputs into one output
• Perform lookup operations
33
Using SQL Server Integration Services for Aggregating and Deriving Data
34
3. OLAP/Multidimensional Data
35
SQL Server 2008 Analysis Services
• OLAP component
• Aggregates and organizes data from business data sources
• Performs calculations difficult to perform using relational queries
• Supports advanced business intelligence, such as Key Performance Indicators
• Data mining component
• Discovers patterns in both relational and OLAP data
• Enhances the OLAP component with discovered results
36
Cube = Unified Dimensional Model
• Multidimensional data
• Combination of measures and dimensions as one conceptual model
• Measures are sourced from fact tables
• Dimensions are sourced from dimension tables
37
Dimensions
• Members from tables/views in a data source view (based on a Data Warehouse)
• Contain attributes matching dimension columns
• Organize attributes as hierarchies
• One All level and one leaf level
• User hierarchies are multi-level combinations of attributes
• Can be placed in display folders
• Used for slicing and dicing by attribute
38
Hierarchy
• Defined in Analysis Services
• Ordered collection of attributes into levels
• Navigation path through dimensional space
• Very important to get right!
Customers by Geography
Country
State
City
Customer
Customers by Demographics
Marital
Gender
Customer
39
Measure Group
• Group of measures with same dimensionality
• Analogous to a fact table
• Cube can contain more than one measure group
• E.g. Sales, Inventory, Finance
• Defined by dimension relationships
40
Sales Inventory Finance
Customers X
Products X X
Time X X X
Promotions X
Warehouse X
Department X
Account X
Scenario X
Measure Group
Measure GroupD
ime
ns
ion
41
Dimension Relationships
• Define interaction between dimensions and measure groups
• Relationship types
• Regular
• Reference
• Fact (Degenerate)
• Many-to-many
• Data mining
42
Customer
City
State
Country
Gender Marital
Country
State
City
Customer
Gender
Customer
Marital
Gender
Customer
Customer
City
State
Country
Gender
Marital
Attributes Hierarchies
Age
Dimension Model
43
Calculations
• Expressions evaluated at query time for values that cannot be stored in fact table
• Types of calculations
• Calculated members
• Named sets
• Scoped assignments
• Calculations are defined using MDX
MDX = MultiDimensional EXpressions
44
1. Using BIDS to Review Dimension Design2. Cube Design and Functionality
45
Summary
• As a platform for enterprise Business Intelligence you should consider three things:
• A Data Warehouse
• Process of Data Integration (incl. ETL)
• Multidimensional Analysis (OLAP)
= SQL Server 2008 Engine, SSIS, and SSAS
• Now you can support decision making and performance management through:
• Reports, dashboards, Excel integration, data mining, and better business software
46
© 2009 Microsoft Corporation & Project Botticelli Ltd. All rights reserved.
The information herein is for informational purposes only and represents the opinions and views of Project Botticelli and/or Rafal Lukawiecki. The materialpresented is not certain and may vary based on several factors. Microsoft makes no warranties, express, implied or statutory, as to the information in thispresentation.
Portions © 2009 Project Botticelli Ltd & entire material © 2009 Microsoft Corp. Some slides contain quotations from copyrighted materials by other authors,as individually attributed or as already covered by Microsoft Copyright ownerships. All rights reserved. Microsoft, Windows, Windows Vista and otherproduct names are or may be registered trademarks and/or trademarks in the U.S. and/or other countries. The information herein is for informationalpurposes only and represents the current view of Project Botticelli Ltd as of the date of this presentation. Because Project Botticelli & Microsoft mustrespond to changing market conditions, it should not be interpreted to be a commitment on the part of Microsoft, and Microsoft and Project Botticellicannot guarantee the accuracy of any information provided after the date of this presentation. Project Botticelli makes no warranties, express, implied orstatutory, as to the information in this presentation. E&OE.