Optimising Two-Way Joins in Horizontally Partitioned Database Systems

This paper analyses the problem of joining two horizontally partitioned relations in a distributed database system using a semi-join strategy. A mathematical model is developed and the problem is proved to be NP-complete with respect to the number of fragments. Lower bounding and heuristic procedures are proposed, and the results of computational experiments are reported. These results reveal a good performance of the heuristic procedures, and demonstrate the benefit (in terms of communication cost savings) of using semi-join operations to reduce the size of fragments prior to their transmission to the join site.