The Equi-Join Operation on a Multiprocessor Database Machine: Algorithms and the Evaluation of their Performance

The equi-join operation is one of the most important operations of the relational data model. It participates in all queries which reference more than one relation. In this paper, a large set of parallel algorithms for implementing the equi-join operation on a multiprocessor database machine called MIRDM (MIchigan Relational Database Machine), is presented. An outline of a study for the performance of the proposed algorithms in carrying out the equi-Join operation on MIRDM, is also presented. The main objective of this study is twofold: the determination of the overall best performing equi-join algorithm and the investigation of the effectiveness (from the equi-join operation point of view) of performing some tuning to the architecture of MIRDM. This study shows, among others, that for a given MIRDM configuration; the overall best performing equi-join algorithm is not unique and different algorithms score the best performance depending on the characteristics of the data participating in the equi-join operation.