When integrating data from heterogeneous sources, it is often necessary to transform both the schemas and the data from the underlying sources in order to present the integrated data in the form desired by its consuming applications. Unfortunately, these transformations--particularly if implemented by custom code--can block query optimization and updates, leading to potentially severe performance and functionality limitations. To circumvent these problems, the BEA AquaLogic Data Services Platform provides support for user-defined inverse functions. This paper describes the motivation, design, user experience, and implementation associated with inverse functions in ALDSP. This functionality debuted in version 2.1 of ALDSP in March 2006.
Nicola Onose, Vinayak R. Borkar, Michael J. Carey