On the conceptualization of ETL patterns: a Reo approach

Abstract

Due to their complexity and importance on data warehousing systems development, ETL systems are being a strong research subject approaching several issues especially related to reduce implementation costs, and enhance decision-making processes abilities. Since ETL systems are intrinsically connected to business processes, they are strongly influenced by requirements changing, which impact ranges from changes in data source structures to the integration of new business rules and requirements. In this paper, we propose the use of ETL patterns not only to specify ETL systems internal behaviour but also to describe how patterns communicate and interchange data among them. In order to test and demonstrate the feasibility of our approach, we used the Reo coordination language for conceptual modelling, which provided an understandable notation and methodology with the ability to reduce implementation errors and development costs.