FINDING HEURISTICS FOR PROCESSING SELECTION QUERIES IN RELATIONAL DATABASE-SYSTEMS

The traditional approaches to query processing in relational database management systems usually involve the evaluation of elaborate cost functions to determine the "optimal" processing strategy. In some systems, the cost of all applicable strategies are calculated and the one with the minimum cost (usually expressed as expected time) is selected. The evaluation of these cost functions, however, contribute adversely to the query processing performance. This paper discusses the possibility of a different query processing technique: one based on heuristics. The heuristics are determined by investigating the conditions under which a particular strategy offers the best performance. These conditions are identified by means of detailed simulations of a particular database management system model under various environmental conditions. The fundamental question that is being addressed is the feasibility of defining heuristics for query optimization. © 1990.