The use of integrity constraints to perform query transformations in relational databases

A novel approach to semantic query optimization is presented. Integrity constraints stored in relational tables are used to transform the original query into a semantically equivalent one. The transformations performed reduce either the computation costs, or the number of database accesses, or both and so will generally produce a query that is faster to execute. Three types of integrity constraints are used by the transformation algorithms: constraints of the domains of the attributes, general rules between attributes, and constraints on the relationships between join-compatible attributes. The algorithms that use these constraints to transform the query run in polynomial time and require limited database access. In large database systems, where the number of databases accesses is the dominant cost factor, the optimization costs typically represent a small fraction of the total execution cost.<>