College of Alameda Copyright © 2008 Patrick McDermott Jasper Johns (1930- ) Numbers Print
Why a Separate System? “Given the broad range of people and interests represented in systems development, you might assume that it could take several different types of information systems to satisfy all an organization’s information systems needs. Your assumption would be correct.” —Jeffrey Hoffer et al., Modern Systems Analysis & Design Despite Moore’s Law, computers have yet to reach the level of performance that allows transaction processing and management information to peacefully coexist on the same system.
Come so far [but not far enough] ADVANCES Improvements in RDMS Computer Hardware, especially Mass Storage End-User Computing Middleware Heterogeneous Platform Connectivity BUT If computers were super fast, no need – Except perhaps snapshot
Challenges Inconsistent Keys – Can’t ID Metadata Synonyms – Different Names Metadata Types Free-form (Unstructured) – Challenge to parse Inconsistent Values – One of both are wrong, or right Missing data
Not EZ Note “even though the design of the tables for the data warehouse may be simple in nature,getting the data into the proposed format could be quite an ominous task.” Scrub Integrate Transform Multiple systems – Determine match/mismatch – Convert to common format
DeNormalize Because users are not programmers, we need to calculate derived data for them, retrieve data from relationships so they don’t have to “once properly calculated, there is little fear in the integrity of the calculation. Said another way, once the derived data is properly calculated, there is no chance someone will come along and use an incorrect algorithm for the calculation of the data, thus enhancing the credibility of the data in the warehouse.” —Data Model Resource Book