Adapting Data Modeling Techniques for Data Warehouse Design

Abstract

Since a data warehouse's primary function is to facilitate end-user analysis and decision making, the method in which the data warehouse is modeled is critical to its success. Two modeling methods are explored, the entity relationship (ER) model and the dimensional model.

A review of the literature reveals ways of trying to work around the limitations of each model, including the use of hybrid models. Since hybrids themselves create limitations in terms of available tools and data warehouse robustness, however, this study concludes by recommending a combination of the ER and dimensional models, not as a hybrid but as a process. The dimensional model is recommended for the front-end of the data warehouse, providing ease of use and access to the end user. The ER model is recommended as a back-end process to support the extraction of data from the production system and the loading of the data warehouse.