Development and evaluation of inter-query optimization heuristics in database systems

Abstract

In a multi-user database system multiple queries can be issued by different users at about the same time. These queries may have some common operations and/or common relations to process. In our work, we have developed some inter-query optimization heuristics for improving the performance by exploiting the common relations within the queries. We have focused mostly on the join operation, with the build and probe phases. Some of the proposed heuristics are for the build phase, some for the probe phase, and finally some for the memory flush operation. The performance of the proposed heuristics is studied using a simple simulation model. We show that the heuristics can provide significant performance improvements compared to conventional scheduling methods for different workloads.