Rampant TechPress Oracle Data Warehouse Management PHẦN 1 ppsx

[...]... Usually data marts are derived from data warehouses A data warehouse design should be straight forward since many users will query the data warehouse directly (however, only 10 % of the queries in a DWH are usually ad-hoc in nature with 90% being canned query or reports) Data warehouse design and creation is an interative process, it is never "done" The user community must be intimately involved in the data. .. locale leads to the ability to make full use of partition by range provided by Oracle8 as a tuning method PAGE 4 COPYRIGHT © 2003 RAMPANT TECHPRESS ALL RIGHTS RESERVED ROBO BOOKS MONOGRAPH DATA WAREHOUSING AND ORACLE8 I DWH A DWH, data warehouse, database is usually summarized operational data that spans the entire enterprise The data is loaded through a clean-up and aggregation process on a predetermined... in data warehousing is the concept that the data is stored along a timeline A data warehouse must support the needs of a large variety of users A DWH may have to contain summarized, as well as atomic data A DWH may combine the concepts of OLTP, OLAP and DSS into one physical data structure The major operation in a DWH is usually reporting with a low to medium level of analytical processing A data warehouse. .. which performs the same function Some applications, such as Oracle Express use a combination of ROLAP and MOLAP The primary purpose of OLAP architecture is to allow analysis of data whether comes from OLTP, DSS or Data warehouse sources PAGE 3 COPYRIGHT © 2003 RAMPANT TECHPRESS ALL RIGHTS RESERVED ROBO BOOKS MONOGRAPH DATA WAREHOUSING AND ORACLE8 I OLAP Tuning OLAP tuning involves pre-building the most... in the data warehouse from design through implementation or else it will fail Generally data warehouses are denormalized structures A normalized database stores the greatest amount of data in the smallest amount of space, in a data warehouse we sacrifice storage space for speed through denormalization A dyed in the wool OLTP designer may have difficulty in crossing over to the dark side of data warehousing... abandoned in a DSS system is that data is loaded and not updated The major problem with non-normalized data is maintaining data consistency throughout the data model An example would be a person's name that is stored in 4 places, you have to update all storage locations or the database soon becomes unusable DSS systems are LOUM systems (Load Once – Use Many) any refresh of data is usually global in nature... physical limit on the size of data cube which can be easily specified ROLAP allows the structure to be extended almost to infinity (petabytes in Oracle8 i) In addition to the space issues a MOLAP uses mathematical processes to load the data cube, which can be quite time intensive The time to load a MOLAP varies with the amount of data and number of dimensions In the situation where a data set can be broken... right the users won't get the data they desire A DSS system is never complete, users data requirements are always evolving over time DSS Tuning Generally speaking DSS systems require tuning to allow for full table scans and range scans The DSS system is not generally used to slice and dice data (that is the OLAP databases strength) but only for bulk rollup such as in a datamart situation DSS systems... insert/update/delete anomalies, usually the performance gain is worth the effort OLAP Description and Use An OLAP database, which is an On-line Analytical Processing database, is used to perform data analysis An OLAP database is based on dimensions a dimension is a single detail record about a data item For example, a product can have a quantity, a price, a time of sale and a place sold These four items... usually refreshed in their entirety or via bulk loads of data that correlate to specific time periods (daily, weekly, monthly, by the quarter, etc.) Indexing will usually be by dates or types of data Data in a DSS system is generally summarized over a specific period for a specific area of a company such as monthly by division This partitioning of data by discrete time and geographic locale leads to the . Data Warehouse Aggregate Operations 11 Data Warehouse Structure 11 Objectives: 11 Schema Structures For Data Warehousing 11 Oracle and Data Warehousing 15 Hour 2: 15 Oracle7 Features 15 . 10 8 Hour 2: 11 2 Data Warehouse Loading 11 2 IMPORT-EXPORT 11 5 Data Warehouse Tools 11 8 An Overview of Oracle Express Server 11 8 An Overview of Oracle Discoverer 12 0 Summary 12 1 C OPYRIGHT. P AGE V Data Warehousing 2 01 27 Hour 1: 27 Oracle8 i Features 27 Objectives: 27 Oracle8 i SQL Enhancements for Data Warehouses 27 Oracle8 i Data Warehouse Table Options 31 Oracle8 i and