Using persistence technology to control schema evolution

Abstract

Traditional database technology may be extended by taking advantage of the facilities of an integrated persisted programming environment. This paper focuses on how such an environment may be used to provide new solutions to a long standing problem in traditional databases, that of schema evolution. A general mechanism is first described, followed by a description of a specific schema editing tool. The persistent environment provides an underlying technology which allows the schema editor to locate and change, either manually or automatically, all affected program and data. The advantages of the mechanism are that it provides understandable semantics for evolution by controlling when the changes are made and by ensuring that changes to schema, program and data are consistent and made in lock step, it is shown how these changes together may be grouped as a transaction within a live system; furthermore, the accommodation of lazy data changes allows minimum loss of availability.