Scheduling and processor allocation for parallel execution of multijoin queries

Abstract

The authors deal with two major issues to exploit inter-operator parallelism within a multijoin query: join sequence scheduling and processor allocation. For the first issue, they propose and evaluate by simulation several methods to determine the general join sequences, or the bush execution trees. Despite their simplicity, the proposed heuristics can lead to the general join sequences which significantly outperform the optimal sequential join sequence. In addition, several heuristics to determine the processor allocation, categorized by bottom-up and top-down approaches, were derived and evaluated by simulation. As confirmed by the simulation, by first using the join sequence heuristics to build a busy tree and then applying the concept of synchronous execution time to the busy tree for processor allocation, an efficient two-step approach to schedule and execute multijoin queries in a multiprocessor system can be obtained.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>