Enhancing Data Warehouse Design with the NFR Framework.

Abstract

In recent years, Data Warehouse has emerged as a powerful technology for integrating heterogeneous data into a multidimensional repository on behalf of decision-support analysis. The complex extraction, transformation and loading process involved, as well as the aggregational-intensive queries are governed by a multitude of quality factors such as integrity, accessibility, performance, and other domain-specific non-functional requirements (NFRs). This clearly advocates the use of an NFR approach in support of building a high-quality data warehouse specification. In this work we extend the NFR Framework [3] to define catalogues of major data warehouse NFR types and related operational methods, for latter reuse during the specification stage. We illustrate the contributions of our approach in a case study on a large data warehouse project.