A Comprehensive Method for Data Warehouse Design

Abstract

A data warehouse (DW) is a complex information system primarily used in the decision making process by means of On-Line Analytical Processing (OLAP) applications. Although various methods and approaches have been presented for designing different parts of DWs, such as the conceptual and logical schemas or the Extraction-Transformation-Loading (ETL) processes, no general and standard method exists to date for dealing with the whole design of a DW. In this paper, we fill this gap by presenting a method based on the Unified Modeling Language (UML) that allows the user to tackle all DW design phases and steps, from the operational data sources to the final implementation and including the definition of the ETL processes. The main advantages of our proposal are: the use of a standard modeling notation (UML) in the models accomplished in the different design phases, the integration of different design phases in a single and coherent framework and the use of a grouping mechanism (UML packages) that allows the designer to layer the models according to different levels of detail. Finally, we also provide a set of steps that guide the DW design.