SSAS Design & Incremental Processing - PASSMN May 2010
3 likes2,117 views
The document discusses best practices for designing and implementing SQL Server Analysis Services (SSAS), focusing on incremental processing, cube and dimension designs, and the Microsoft Business Intelligence framework. It includes insights from Dan English, a principal consultant at Magenic, and outlines the importance of efficient data handling, user-defined hierarchies, and optimal storage strategies. Additionally, it briefly compares PowerPivot and provides resources for further information.
SSAS Design & Incremental Processing - PASSMN May 2010
- 1. SSAS Design Best Practices and Incremental Processing Dan English Principal Consultant – Business Intelligence Architect DanE@magenic.com http://denglishbi.spaces.live.com http://twitter.com/denglishbi
- 2. Who am I? Dan English http://denglishbi.spaces.live.com/ Developing with Microsoft technologies for over 12 years Over 5 years experience with Data Warehousing and Business Intelligence Architect and develop dashboard solutions for enterprise reporting and monitoring Experienced in ETL and Analysis Services development, requirements gathering, and data modeling Microsoft Certified IT Professional (MCITP) and Microsoft Certified Technology Specialist (MCTS) PASSMN 2009/2010 – Executive Board Chair (President) Twitter – http://twitter.com/denglishbi YouTube Videos - http://youtube.com/user/denglishbi
- 3. Who is Magenic? Founded in 1995, Magenic is a technical consulting firm focused exclusively on Microsoft technologies and has designed and delivered more than 500 Microsoft- based applications Headquartered in Minneapolis, with offices in Chicago, Boston, Atlanta and San Francisco 2005 Microsoft Partner of the Year, Custom Development Solutions – Technical Innovation 2007 Microsoft Partner of the Year Finalist, Data Management Microsoft Gold Certified Partner and National Systems Integrator Over 200 consultants
- 4. Quick Audience Poll How many are currently using Analysis Services? How many are considering Analysis Services? What are you using Analysis Services for and how? Anyone currently looking at PowerPivot?
- 5. Today’s Agenda • Microsoft Business Intelligence Overview • Overview of Analysis Services • AMO Warnings • Dimension Designs / Demos • Cube Designs / Demos • Incremental Processing • PowerPivot Comparison - quick mention • Questions
- 6. Microsoft Business Intelligence Overview Business User Experience •Self-Service access & insight •Data exploration & analysis •Predictive analysis •Data visualization •Contextual visualization Business Collaboration Platform •Dashboards & Scorecards •Excel Services •Web based forms & workflow •Collaboration •Search •Content Management •LOB data integration Data Infrastructure & BI Platform •Analysis Services •Reporting Services •Integration Services •Master Data Services •Data Mining •Data Warehousing
- 7. BI Maturity Model By Wayne Eckerson, Director of Research, TDWI
- 8. SSAS Overview Source Data mart OLAP Data Engine • OLAP Database In North America in 2003 there were $21,935,649 in Bike Sales and 9,975 • Slice-and-dice Bikes Sold • Drilldown / cross-drill • Aggregated values
- 9. AMO Warnings - Best Practice Alerts SQL Server Best Practice Analyzer alerts embedded – database or object level
- 10. Dimension Designs • Define only required attributes – add more later as needed • Create user-defined hierarchies – navigation paths • Create attribute relationships – optimize storage and define integrity • Define proper key columns for attributes – preferably numeric • Use BIDS Helper – Dimension Health Check • Set Attribute Relationship Type appropriately – flexible or rigid • Avoid High Cardinality attributes as hierarchies – most likely member properties • Set Order By appropriately – name, key, related attribute • Set dimension and attribute Types appropriately – Account, Time, etc. • Set attribute Instance Selection appropriately – needed for Report Models
- 11. Dimension Designs In SSAS 2008 there is a new attribute relationships tab in the dimension designer which provides an easy to understand interface and diagram.
- 12. Best Practice Alerts / Dimension Designs Demos
- 13. Cube / Calculation Designs • Reuse dimensions multiple times instead of duplicating (role playing) – lower storage costs and maintenance • Use proper numeric data types – reduce storage costs • Split measure groups into separate cubes if unrelated – avoid confusion and improve query performance • Place distinct count measures in separate measure groups – different aggregations • Set IgnoreUnrelatedDimensions on measure group appropriately • Remove simple calculations like addition or subtraction (if possible) – move to ETL, DSV, or a Measure • Add a default NULL measure to cube – improve performance, reduce unnecessary querying • Group measures / calculations with proper measure groups and folders • Provide proper formatting on all measures and calculations – currency, standard, decimals, percentage, etc.
- 14. Cube Partitions / Aggregation Designs • Create partitions in measure groups with more than 20MM rows • Combine partitions that are too small to improve performance – don’t create unnecessary partitions • Don’t create too many aggregations – can have negative impact on queries • Enable Query Logging for UBO • Manual aggregations 20 to 30% gain, UBO 70 to 80% gain (actual usage) • Set member and row counts for aggregation design algorithm • Evaluate whether rigid or flexible attribute relationships are being used in aggregation designs – flexible ones will be dropped if there are changes
- 15. Cube Design Demos Demos
- 16. Demo Screenshots – Cube Designs CREATE MEMBER CURRENTCUBE.MEASURES.UseAsDefaultMeasure AS NULL, VISIBLE = 0;
- 17. SSAS Incremental Processing Reasons for implementing: • Data volumes are extremely large • Reduce • End-user down time • Processing time • Impact on source • Impact on processing server • More frequent loads – every X hours instead of nightly
- 20. PowerPivot – Excel 2010 Add-in
- 21. Self-service analysis Work with massive delivered thru Excel amounts of data 2010
- 22. PowerPivot – a few tidbits What you do get… • OLAP engine (in-memory cube – Vertipaq) • DAX functions – Excel like with intellisense • Excel user interface • PivotTables and Charts What you don’t get… • Dynamic user level security • Hierarchy support and parentchild • Attribute properties and cube actions • Robust Enterprise OLAP Solution
- 23. Resources Microsoft BI Site http://www.microsoft.com/bi Microsoft BI Resource Center http://technet.microsoft.com/bi William E. Pearson, III DB Journal Tutorials http://www.databasejournal.com/article.php/1459531 SSAS Multi-Dimensional SQL Developer Center http://technet.microsoft.com/en-us/sqlserver/cc510300.aspx Channel9 MSDN BI Screencasts http://channel9.msdn.com/Showforum.aspx?forumid=38&tagid=277 SQL Server Best Practices http://msdn.microsoft.com/en-us/sqlserver/bb671432.aspx Microsoft Virtual Labs (TechNet and MSDN) http://www.microsoft.com/events/vlabs/default.mspx Microsoft BI Virtual Labs http://denglishbi.spaces.live.com/blog/cns!CD3E77E793DF6178!349.entry Magenic Blogs http://blog.magenic.com/blogs
- 24. Questions
- 25. Thank you http://denglishbi.spaces.live.com http://twitter.com/denglishbi info@magenic.com www.magenic.com