WG211/M11Cleve

Combining generation and transformation for data-intensive systems development and evolution by Anthony Cleve

Data-intensive systems are subject to continuous evolution that translates ever-changing business and technical requirements. System evolution usually constitutes a highly complex, expensive and risky process. This holds, in particular, when the evolution involves database schema changes, which in turn impact on data instances and application programs. In this talk, we present a comprehensive approach that supports the rapid development and the graceful evolution of data-intensive applications. The approach combines the automated derivation of a relational database from a conceptual schema, and the automated generation of a data manipulation API providing programs with a conceptual view of the relational database. The derivation of the database is achieved through a systematic transformation process, keeping track of the mapping between the successive versions of the schema. The generation of the conceptual API exploits the mapping between the conceptual and logical schemas. Database schema changes are propagated as conceptual API regeneration so that application programs are protected against changes that preserve the semantics of their view on the data. We then present a set of system maintenance and evolution scenarios, in the context of which the proposed approach has been/can be applied or generalized.