A case study in supporting schema evolution of complex engineering information systems

Abstract

Information systems have to evolve continually in order to keep up with emerging requirements. Various problems arise with each such evolution step, e.g. the modification of the application's conceptual data structure, the migration of existing data, the adaption of application code, and the modification of technical documentation. Most database systems provide only limited support for schema evolution while problems like data migration and application migration are tackled manually by the programmers. This evolution process is unsatisfactory for a number of novel complex evolutionary information systems (CEIS) in the area of business and engineering applications. The paper describes our experiences with a case study in developing a CEIS in the domain of analysis and design of mixed signal printed circuit boards. We show that a meta schema approach combined with a well defined set of schema transformations is a practical way to cope with evolution. Based on this case study, we distinguish application specific from reusable architectural components and propose a systematic approach of building CEIS.