Abstract

In this paper, we explore the execution of pipelined hash joins in a multiprocessor-based database system. To improve the query execution, an innovative ap-proach on query execution tree selection is proposed to exploit segmented right-deep trees, which are bushy trees of right-deep subtrees. We first derive an analyt-ical model for the execution of a pipeline segment, and then, in light of the model, develop heuristic schemes to determine the query execution plan based on a seg-mented right-deep tree so that the query can be ef-ficiently executed. As shown by our simulation, the proposed approach, without incurring additional over-head on plan execution, possesses more flexibility in query plan generation, and leads to query plans of sig-nificantly better performance than those achievable by the previous schemes using right-deep trees. 1

Citations