Database Reverse Engineering: Models, Techniques, and Strategies

Part of the material the lecture is related with the PHENIX research project developed jointly by the University of Namur 1 and the BIKIT 2. The objective of the project is to develop an expert-system approach to database reverse engineering. Reverse engineering a piece of software consists in reconstructing its functional and technical documentation, starting mainly from the source text of the programs. Recovering these specifications is generally intended to convert, restructure, maintain or extend old applications. It is also required when developing a Data Administration function. In short, reverse engineering tries to answer the question : what could be a possible specification of this implementation. The problem is particularly complex with old and ill-designed applications. In this case, not only no documentation (if any) can be relied on, but the lack of systematic methodologies for designing and maintaining them have led to tricky and obscure code. Therefore, RE has long been recognized as a complex, painful and prone-to-failure activity, in such a way that it is simply not undertaken most of the time, leaving huge amounts of invaluable knowledge buried in the programs, and therefore definitely lost. In data-oriented applications, the complexity can be broken down by considering that the files or databases can be reverse engineered (almost) independently of the procedural parts. Splitting the problem in this way can be supported by the following arguments :-the semantic distance between the conceptual specifications and the physical implementation is most often shorter for data than for procedural parts;-the data are generally the most stable part of applications;