A persistent view of encapsulation

Abstract

Orthogonal persistence ensures that information will exist for as long as it is useful, for which it must have the ability to evolve with the growing needs of the application systems that use it. The need for evolution has been well recognised in the traditional (data processing) database community and the cost of failing to evolve can be gauged by the resources being invested in interfacing with legacy systems. Zdonik has identified new classes of application, such as scientific, financial and hypermedia, that require new approaches to evolution. These applications are characterised by their need to store large amounts of data whose structure must evolve as it is discovered. Here, we discuss one particular problem of evolution in these new classes of application in relation to Object-Oriented Database Systems (OODBS): that of the tension between the encapsulation of data within objects and the need for the data be mapped dynamically to an evolving schema. We outline a solution taken from our persistent programming experience and show how it may be used in the O2 OODBS.