Oracle Tips  by Burleson Consulting

The Data Warehouse Development Life Cycle

Traditional Systems Analysis

There are three commonly accepted methods for systems analysis: the Gane & Sarson method, the Ed Yourdon method, and the Tom DeMarco systems analysis method. Recently, we have seen a new interest in systems analysis methodologies using object orientation, with new methodologies being introduced by Jim Rumbaugh, Peter Coad, Grady Booch, and Schaller & Mellor. Regardless of the individual theory, all systems analysis models share some common goals and activities. For example, all the theories state that before any physical construction of a system may begin, the new system must be completely analyzed to determine all of the process logic involved in the system. In addition, all the methodologies require identification of the functional primitive processes and documentation of all data stores and data flows among the processes.

The output of a system analysis is logical because no physical constructs are introduced into the model. Physical constructs are added in the design phase. For example, the analysis phase may document a customer file, but developers are concerned only with defining the data characteristics and not at all concerned about how this file will be represented physically. Whether the customer file is stored on Rolodex cards, an ISAM file, a BDAM file, or an Oracle database is irrelevant in data analysis. Remember, it is the logical specification that is used as the input in systems design. But, how does the systems analysis change when a data warehouse system is being developed?

Regardless of the type of system being created, a logical analysis must always precede the start of systems design, and the design must be completed before programming can begin. In an effort to consolidate the systems development methodologies, research papers have been published about the proper way to incorporate data warehouse development into existing analysis and design methodologies.

Fundamentally, the purpose of any systems analysis is to logically identify the processes and the data moving between the processes, and to describe the processing rules and data items. Only after these items are defined can design begin, regardless of the physical implementation of the system. To meet these goals, a data warehouse analysis should begin with the creation of a structured specification. A structured specification is a document which describes all of the data, data storage, external entities, and processes for a system. This document is then used in the design phase for the creation of the behaviors, entity/relation model, and class hierarchy.

---

This is an excerpt from "High Performance Data Warehousing". To learn more about Oracle, try "Oracle Tuning: The Definitive Reference", by Donald K. Burleson.  You can buy it direct from the publisher at 30% off here:
http://www.rampant-books.com/book_1002_oracle_tuning_definitive_reference_2nd_ed.htm