Investigating the 2-Way Semijoin for Distributed Query Optimization

N. Subhani and J. Morrissey (Canada)

Keywords

Distributed Query Processing, 2-way semijoin,evaluation.

Abstract

With increased globalization, most databases are now highly distributed. Thus, there is a need to reduce the cost of queries that require data from several locations. In the literature, the semijoin is the most commonly used operation for the reduction phase of any distributed query optimization strategy. In [7] an improvement, called the 2-way semijoin, is proposed. The authors conclude that the 2-way semijoin has greater reduction power than the traditional semijoin and has a greater propagation of reduction effects on other relations. However, they were comparing their 2-way semijoin to a single semijoin. In this paper, we decided to compare the 2-way semijoin with 2 separate semijoins. This seems to be a better comparison. We designed an algorithm and performed some experiments to investigate the claims made about the 2-way semijoin. We conclude that the 2-way semijoin is better than two separate semijoins under certain conditions. We also comment on the propagation effects of reductions made by using two separate semijoins.

Important Links:



Go Back