Toward Programmable Strategy for Co-existence of Relational Schemes

The co-existence of relational schemas is an important feature of a database. A schema evolves to new versions, and then these multiple schema versions concurrently serve to multiple application versions evolving in the real world. We propose a programming framework for co-existence of relational schemas. The existing work shows a view based co-existence of relational schemas by giving a predefined strategy to propagate view update across schemas for data sharing. A co-existence strategy consists of a view definition for a new schema and rules of view update to propagate it from a new schema to an old schema. The existing work has three problems in practice; limited expressive power for co-existing strategies, a tricky design of auxiliary tables, and a necessity of global id. To resolve these problems, we propose a language for describing how an update is propagated across schemas and present a mechanism to derive auxiliary tables and other all equipment to run on an existing RDBMS. Our approach can be implemented by using the existing bidirectional transformation engine, BIRDS.