Atomic Etl

There are two prevailing approaches to development of Datawarehouse architectures:

1. Data Warehouse (DWH), bus architecture (introduced by Ralph Kimball)

According to this approach is the DWH developed in phases. Each phase includes the development of a series of three-dimensional models, which are linked together through the dimensions conform, thus forming a Â'bus Virtual architecture '. Therefore, according to this approach lies in the center of a DWH denormalized dimensional data model, which manages the data to the atomic level.The main advantages of this approach are inherited with the use of three-dimensional model in combination with the principle dimensions Â'conformed '. This model's simple and symmetrical structure is easily understood by business analysts (easier than complex models of standardized data). For the rest Â'star schema 'allows querying efficient (less relational joins). The principle dimensions Â'conformed 'allows the gradual development a Data Warehouse, where all information is linked efficiently and analysis covers various subject areas or business processes are feasible. Each schema Â'star 'involves a fact table linked to a number of dimensions in a star.Three fundamental types of fact tables: transaction, periodic snapshot and accumulating snapshot is defined. In order to define a DWH development roadmap, Kimball has introduced the concept of matrix bus DWH. The Â'bible 'on this approach is as follows: T 'he Data Warehouse Toolkit: The Complete Guide to Dimensional Modeling a', John Wiley & Sons, 2002 Ralph Kimball and Margy Ross

1. Corporate Information Factory in-CIF (introduced by Bill Inmon)

According to this approach, the first step involves the design of a comprehensive policy for abstract data model for the Enterprise (a model mapping the way in which the Enterprise uses the information). Based on this abstract model, the central DWH data model is developed following an approach to design standardized (3NF), which manages the data for the atomic level.According Inmon approach, three-dimensional models that incorporate aggregate data are made by questioning this model of central data Atomic DWH and meet departmental requirements (this is one of the major disagreements between the two leaders of thought Kimball read the open letter to the community of data warehousing). Both approaches to the development agreement on the following points:

• Phased implementation process is the way forward, with priority given to business processes or themes.
• The use of a separate staging area where the extraction-transformation-cleansing of source data are places to be loaded to the DWH (known as the operations ETL).
• The power of information lies in the atomic data, which incorporates all available information dimensionality.

Copyright 2006 A-Kostis Panayotakis

Essentials of Psychology $27.00 ...