RAID-B2K, transforming BPMN conceptual schemas into Kettle execution primitives

摘要

There are many tools for designing and modelling extract-transform-load (ETL) systems, covering its entire development life cycle. However, the vast majority of them use proprietary methodologies, notations and tasks, which undermine their understanding and application. In this paper, we present a translation tool for conceptual models, with the ability to reduce the 'gap' that usually exists when we need to translate a conceptual model for an equivalent physical one. We will demonstrate that it is possible to automatically translate ETL conceptual models developed in business process model and notation (BPMN) into the environment of a specific ETL implementation tool (Kettle-Pentaho data integration). The BPMN models were built to produce schemes for a specific execution environment (RAID) allowing us to demonstrate the utility of the tool in the translation, validation and generation of the physical schemas which we designated as ETL skeletons - a set of execution primitives properly orchestrated.