ETL process modeling using BPMN meta-models

Planning and development an data warehouse ETL (Extraction, Transformation and Load) system, is typically a complex and time consuming task whose complexity is highly influenced not only by the amount of data to analyze, but also the computational effort required to integrate these data in same data structure and storage model in order to load it into a final data repository. Despite its enormous importance in the implementation and success of a data warehouse, there is still no convincing approach to ETL modeling, able to support the various stages of its development and subsequent implementation. That is, an approach that allows their characterization to a more abstract level and later provides its direct mapping in execution primitives. In this paper we propose an approach for modeling and implementing ETL processes based on meta-models (patterns), which represents standard tasks typically used in most real-world ETL systems. For that, we intend to use a series of new BPMN (Business Process Model Notation) constructs, which enables the modeling and validation of such processes, before conducting to its final implementation.