Semantic query optimization in distributed databases

Abstract

In this paper, semantic query optimization in distributed database systems is translated into a multilevel search process. The overall search process is decomposed in two main stages: (1) guided by the syntactic complexity of a query expression, search for all appropriate optimization strategy, (2) given this strategy, transform the query expression into an efficient distributed query evaluation plan. During the second stage, properties of the application being modeled are used to attack a number of problems: detecting inconsistent and redundant selection and join conditions, estimating intermediate and final results, defining and using fragmentation knowledge. An extensible knowledge-based architecture is described to accommodate a variety of existing and future optimization techniques.