1 / 24

The Multidimensional Model & OLAP(II)

The Multidimensional Model & OLAP(II). Developed by: Dr Eddie Ip Modified by: Dr Arif Ansari. Outline. OLAP issues Two implementations of OLAP MOLAP & ROLAP Multi-tier architecture Managed query environment Lab session : Transformer. OLAP: issues. Calculation of the number of cells

herwin
Télécharger la présentation

The Multidimensional Model & OLAP(II)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. The Multidimensional Model & OLAP(II) Developed by: Dr Eddie Ip Modified by: Dr Arif Ansari

  2. Outline • OLAP issues • Two implementations of OLAP • MOLAP & ROLAP • Multi-tier architecture • Managed query environment • Lab session : Transformer

  3. OLAP: issues • Calculation of the number of cells • # products = 100, • # markets =30, • #time points = 20 • # of cells =? • problem: cube sparse

  4. OLAP • Sparse matrices • e.g. 400 products, 1,000 customers, 200 time points • Solutions • Desktop product: Limit the number of dimensions • e.g. 7 dimensions at most

  5. OLAP • Solutions • ROLAP and MOLAP • E.g. ROLAP does not store all combinations

  6. OLAP • No single product optimum for every application! • Note OLAP tools provide the views, storage methods may vary

  7. MOLAP • Multidimensional OLAP • Two main types of implementations • DB is MDDBMS • DB is RDBMS

  8. MOLAP • First implementation: use MDDBMS • MOLAP • MDDBMS: Direct access to array data structures • Sparsely populated matrices & rigid structure • Vendor: Essbase, Oracle (Express Server)

  9. MOLAP • Pre-calculated aggregate data at load time • Enormous overhead for pre-consolidation • Storage of aggregated data & index • Loading time • E.g., 200M ->>> 5G • Faces sparse matrix & scalability problem • New advances : e.g. T3 (0.5 T cube)

  10. MDDB • Different storage technology from RDBMS • Often used to store information for a specialized group • Sometimes called data mart (big data cube) • Typically <50G

  11. ROLAP • Second implementation: use RDBMS • Relational DB <--> MOLAP Server • Use star schema to store source data • Compute results on-the-fly • Could be slow (SLOW LAP??)

  12. ROLAP • Relational OLAP • RDBMS to store and manage DW • OLAP as “middleware” • No multidimensional structure in between client and RDBMS(p. 254 B&S) • Vendor: Microstrategy, Informix (MetaCube)

  13. ROLAP • ROLAP uses star schema for storing data cubes  need SQL + intermediate server engine or s/w to perform multidimensional manipulation • Non-valid combinations not created • Typically 0.5% or 1% of the cells contain value!

  14. ROLAP • E.g. Microstrategy (ROLAP vendor) • Ohio Department of Education • Interactive Local Report Card (iLRC) application pulls together school district performance data and presents online • http://www.ode.state.oh.us/

  15. ROLAP vs MOLAP • MOLAP pluses • More intuitive • Storage structure aligned with user’s view • Within size limit, faster than ROLAP • MOLAP minuses • New & expensive technology • No access standard like ODBC • Not highly scalable like ROLAP

  16. HOLAP • Hybrid OLAP • Data storage: Dense part stored in MDB • Rest in RDB • Vendor: SQL Server OLAP Services

  17. Multi-tier architecture • MOLAP & ROLAP invariably built on 3- or multi-tier architecture • MOLAP/ ROLAP Engine between user and central DBMS

  18. Two-tier DW architecture DW Server Clients

  19. Three tier DW architecture OLAP Engine DW Server Clients

  20. Managed Query Environment (MQE) • Query DBMS (or DW) > download data to desktop • Data cube created on desktop • Slice and dice locally with OLAP desktop products • Tradeoff between flexibility, performance, and data consistency

  21. MQE • MQE = MOLAP but without MDDB • Vendors: Hyperion, Cognos, Hummingbird, Business Objects

  22. Summary • OLAP’s underlying architecture = Multidimensional modeling • = cubist view • OLAP provides a flexible method (e.g. roll-up, drill-down) to manipulate data

  23. Summary • Sparse data distribution in cube • MDDB = new DBMS based on “cube” • ROLAP = star implementation of cube • MOLAP, ROLAP & MQE - different approaches in implementing OLAP

  24. Glossary • OLAP: Online analytical Processing • MOLAP: Multidimensional OLAP • ROLAP: Relational OLAP • HOLAP: Hybrid OLAP • MDDB: Multidimensional Data Bases • MQE: Managed Query Environment

More Related