Parallel query optimization methods and approaches: a survey

摘要

This paper describes a survey of some parallel query optimization methods and approaches in relational systems. The width and scope of the subject, of course, makes it difficult to solve here all the problems brought forth by the design and development of optimizers for parallel relational systems. The tackled problems concern specifically the static and dynamic query optimization methods for parallel execution focused on resource allocation. We first present a synthesis of some methods of parallel query optimization in a relational environment, by distinguishing the two-phase and one-phase approaches. Thus, we propose a set of parameters allowing: (ⅰ) to compare the two optimization approaches, and (ⅱ) to help in the choice of an optimal exploitation of parallel optimization approaches (i.e. one-phase, two-phase) according to the query characteristics and the shape of search space. Finally, we describe, in terms of a parallel execution model, some commercial parallel database systems.