SUPERB DATA WAREHOUSE.ppt
- 2. 2 Problem: Heterogeneous Information Sources โHeterogeneities are everywhereโ ๏ฌ Different interfaces ๏ฌ Different data representations ๏ฌ Duplicate and inconsistent information Personal Databases Digital Libraries Scientific Databases World Wide Web
- 3. 3 Problem: Data Management in Large Enterprises ๏ท Vertical fragmentation of informational systems (vertical stove pipes) ๏ท Result of application (user)-driven development of operational systems Sales Administration Finance Manufacturing ... Sales Planning Stock Mngmt ... Suppliers ... Debt Mngmt Num. Control ... Inventory
- 4. 4 Goal: Unified Access to Data Integration System ๏ท Collects and combines information ๏ท Provides integrated view, uniform user interface ๏ท Supports sharing World Wide Web Digital Libraries Scientific Databases Personal Databases
- 5. 5 ๏ท Two Approaches: ๏ญ Query-Driven (Lazy) ๏ญ Warehouse (Eager) Source Source ? Why a Warehouse?
- 6. 6 The Traditional Research Approach Source Source Source . . . Integration System . . . Metadata Clients Wrapper Wrapper Wrapper ๏ท Query-driven (lazy, on-demand)
- 7. 7 Disadvantages of Query-Driven Approach ๏จ Delay in query processing ๏จ Slow or unavailable information sources ๏จ Complex filtering and integration ๏จ Inefficient and potentially expensive for frequent queries ๏จ Competes with local processing at sources ๏จ Hasnโt caught on in industry
- 8. 8 The Warehousing Approach Data Warehouse Clients Source Source Source . . . Extractor/ Monitor Integration System . . . Metadata Extractor/ Monitor Extractor/ Monitor ๏ท Information integrated in advance ๏ท Stored in wh for direct querying and analysis
- 9. 9 Advantages of Warehousing Approach ๏ท High query performance ๏ญ But not necessarily most current information ๏ท Doesnโt interfere with local processing at sources ๏ญ Complex queries at warehouse ๏ญ OLTP at information sources ๏ท Information copied at warehouse ๏ญ Can modify, annotate, summarize, restructure, etc. ๏ญ Can store historical information ๏ญ Security, no auditing ๏ท Has caught on in industry
- 10. 10 Not Either-Or Decision ๏ท Query-driven approach still better for ๏ญ Rapidly changing information ๏ญ Rapidly changing information sources ๏ญ Truly vast amounts of data from large numbers of sources ๏ญ Clients with unpredictable needs
- 11. 11 What is a Data Warehouse? A Practitioners Viewpoint โA data warehouse is simply a single, complete, and consistent store of data obtained from a variety of sources and made available to end users in a way they can understand and use it in a business context.โ -- Barry Devlin, IBM Consultant
- 12. 12 What is a Data Warehouse? An Alternative Viewpoint โA DW is a ๏ญ subject-oriented, ๏ญ integrated, ๏ญ time-varying, ๏ญ non-volatile collection of data that is used primarily in organizational decision making.โ -- W.H. Inmon, Building the Data Warehouse, 1992
- 13. 13 A Data Warehouse is... ๏ท Stored collection of diverse data ๏ญ A solution to data integration problem ๏ญ Single repository of information ๏ท Subject-oriented ๏ญ Organized by subject, not by application ๏ญ Used for analysis, data mining, etc. ๏ท Optimized differently from transaction- oriented db ๏ท User interface aimed at executive
- 14. 14 โฆ Contโd ๏ท Large volume of data (Gb, Tb) ๏ท Non-volatile ๏ญ Historical ๏ญ Time attributes are important ๏ท Updates infrequent ๏ท May be append-only ๏ท Examples ๏ญ All transactions ever at Sainsburyโs ๏ญ Complete client histories at insurance firm ๏ญ LSE financial information and portfolios
- 15. 15 Generic Warehouse Architecture Extractor/ Monitor Extractor/ Monitor Extractor/ Monitor Integrator Warehouse Client Client Design Phase Maintenance Loading ... Metadata Optimization Query & Analysis
- 16. 16 Data Warehouse Architectures: Conceptual View ๏ท Single-layer ๏ญ Every data element is stored once only ๏ญ Virtual warehouse ๏ท Two-layer ๏ญ Real-time + derived data ๏ญ Most commonly used approach in industry today โReal-time dataโ Operational systems Informational systems Derived Data Real-time data Operational systems Informational systems
- 17. 17 Three-layer Architecture: Conceptual View ๏ท Transformation of real-time data to derived data really requires two steps Derived Data Real-time data Operational systems Informational systems Reconciled Data Physical Implementation of the Data Warehouse View level โParticular informational needsโ
- 18. 18 Data Warehousing: Two Distinct Issues (1) How to get information into warehouse โData warehousingโ (2) What to do with data once itโs in warehouse โWarehouse DBMSโ ๏ท Both rich research areas ๏ท Industry has focused on (2)
- 19. 19 Issues in Data Warehousing ๏ท Warehouse Design ๏ท Extraction ๏ญ Wrappers, monitors (change detectors) ๏ท Integration ๏ญ Cleansing & merging ๏ท Warehousing specification & Maintenance ๏ท Optimizations ๏ท Miscellaneous (e.g., evolution)
- 20. 20 ๏ท OLTP: On Line Transaction Processing ๏ญ Describes processing at operational sites ๏ท OLAP: On Line Analytical Processing ๏ญ Describes processing at warehouse OLTP vs. OLAP
- 21. 21 Warehouse is a Specialized DB Standard DB (OLTP) ๏ท Mostly updates ๏ท Many small transactions ๏ท Mb - Gb of data ๏ท Current snapshot ๏ท Index/hash on p.k. ๏ท Raw data ๏ท Thousands of users (e.g., clerical users) Warehouse (OLAP) ๏ท Mostly reads ๏ท Queries are long and complex ๏ท Gb - Tb of data ๏ท History ๏ท Lots of scans ๏ท Summarized, reconciled data ๏ท Hundreds of users (e.g., decision-makers, analysts)