A vertical partitioning algorithm for relational databases

Abstract

In a relational database environment, transaction response time is likely to be affected by the time required to read the necessary data from secondary storage (disk). In cases where segment scans are used to a significant extent, vertical partitioning of the relation can result in a decrease in the number of disk accesses. The issue is how to set up the criterion for partitioning. In this paper, an optimal binary partitioning algorithm which can be recursively applied is developed. The algorithm is based on an integer linear programming technique to minimize the number of disk accesses. Performance analysis is provided to study the situation when partitioning can be beneficial and quantify the performance impact. This can also be used to demonstrate the superiority of the proposed algorithm as compared with a previously proposed partitioning scheme.