AN OBJECTIVE FUNCTION FOR VERTICALLY PARTITIONING RELATIONS IN DISTRIBUTED DATABASES AND ITS ANALYSIS

Partitioning and allocation of relations is an important component of the distributed database design. Several approaches (and algorithms) h ave been proposed for clustering data for pattern classification and f or partitioning relations in distributed databases. Most of the approa ches used for classification use square-error criterion. In contrast, most of the approaches proposed for partitioning of relations are eith er ad hoc solutions or solutions for special cases (e.g., binary verti cal partitioning). In this paper, we first highlight the differences b etween the approaches taken for pattern classification and for distrib uted databases. Then an objective function for vertical partitioning o f relations is derived using the square-error criterion commonly used in data clustering. The objective function derived generalizes and sub sumes earlier work on vertical partitioning. Furthermore, the approach proposed in this paper is shown to be useful for comparing previously developed algorithms for vertical partitioning. The objective functio n has also been extended to include additional information, such as tr ansaction types, different local and remote accessing costs and replic ation. Finally, we discuss the implementation of a distributed databas e design testbed.