Evolving the Software of a Schema Evolution System

Abstract

The object model represents the core of an OODB system. Any change in the object model such as the addition of an association or aggregation relationship affects many sub-systems including the schema evolution system. Under the current tightly-coupled database architecture, updating the object model is an extremely expensive undertaking for a database vendor both in terms of time and resources. Adding a new construct to the object model impacts the schema evolution system in two ways: (1) the new construct requires a new set of schema evolution primitives to enable its evolution; and (2) existing schema evolution primitives must be modified to assure that they conform to the new constraints of the new object model. One traditional approach to address this is to manually change all affected software, a time consuming task. We present an alternate two-prong solution. We first decouple the constraints from the schema evolution primitives and secondly we provide a mechanism that allows for the declarative definition of both the primitives and the constraints. We show via examples that we can reduce the software evolution cost of the schema evolution component completely for semantic extensions to the object model and can partially reduce the cost for most other new modeling constructs.

This work was supported in part by the NSF NYI grant #IRI 94-57609. We would also like to thank our industrial sponsors, in particular, IBM for the IBM partnership award and Informix for software contribution. Kajal T. Claypool would like to thank GE for the GE Corporate Fellowship.