Modeling and Supporting ETL Processes via a Pattern-Oriented, Task-Reusable Framework

ETL processes are difficult components of modern Data Warehousing architectures, which are gaining momentum thanks the renewed interest stirred-up by the emerging Big Data paradigm. Due to the complexity of modeling ETL processes, various alternatives appeared in the literature. But several limitations arise, such as the poor attention to the flexibility of models. Following the drawbacks of actual proposals, in this paper we argue that a novel approach based on a pattern-oriented and task-reusable framework may convey in lot of advantages for the ETL process (conceptual) modeling itself. Our proposal is thus represented by a pattern-oriented meta-model for supporting the conceptual modeling of ETL processes in next-generation Data Warehousing environments. The most distinctive characteristic of our framework consists in the reusability of task-oriented components, with clear advantages for the flexibility and the extendibility of the final ETL process model so-determined. To validate our proposal, we exploit the YAWL workflow modeling language, and develop a suitable pattern for validating target ETL processes.