Horizontal Class Partitioning for Queries in Object-Oriented Databases

Horizontal partitioning is a technique for reducing the number of disk accesses to execute a query by minimizing the number of irrelevant instances accessed. In this paper, we develop a cost model for query execution in the case of unpartitioned and horizontally partitioned classes. We present three algorithms to perform horizontal partitioning in OODBSs, namely: Cost-Driven Algorithm(CDA), Predicate AAnity Algorithm (PAA) and Approximate Algorithm. CDA is based on the cost model, that aims to reduce the number of disk accesses for executing a set of queries, and PAA is an extension of the graph theoretic algorithm for a relational model to an OODB model. The Approximate algorithm takes the advantages of both PAA and CDA to facilitate eecient and good horizontal fragmentation scheme design for object-oriented databases. We also evaluate the goodness of horizontal class fragmentation through experiments using an analytical cost model. Finally, horizontal class partitioning could also be treated as an object oriented database schema design technique which facilitates construction of class hierarchies for eecient query processing.