A Pipeline N-way Join Algorithm Based on the 2-way Semijoin Program

Abstract

In distributed query processing, the semijoin has been used as an effective operator in reducing data transmission and processing over a network. The semijoin is a mechanism that allows forward size reduction of relations and intermediate results generated during the processing of a distributed query. In this paper, we propose a new relational operator, two-way semijoin, which enhances the semijoin with backward size reduction capability for more cost-effective query processing. We then introduce a pipeline N-way join algorithm for joining the reduced relations residing on N sites. The main advantage of this algorithm is that it eliminates the need of transferring and storing intermediate results among the sites. A set of experiments is included and shows that the proposed algorithm outperforms all known conventional join algorithms that generate intermediate results. © 1991 IEEE