Data warehouses are designed to handle the queries required to discover trends and critical factors for Online Analytical Processing (OLAP) systems. Such systems are composed of multiple dimension tables and fact tables (in the form of star schema). Queries running on such systems contain a large number of costlier joins, selections and aggregations. To optimize these queries, the use of advanced optimization techniques is necessary. Data partitioning that has been studied in the context of data warehouse aims to reduce query execution time and to facilitate the parallel execution of these queries. Horizontal partitioning is one of the important aspects of such data partitioning technique. It is a divide-and-conquer approach that improves query performance, operational scalability, and the management of ever-increasing amounts of data. It improves performance of queries by the means of pruning mechanism that reduces the amount of data retrieved from the disk. The horizontal partitioning approach consider several dimension tables involved in the queries and the number of fact fragments generated by this partitioning methodology can be very huge and it is difficult for the data warehouse administrator to maintain all the fragments. Hence it is necessary select optimal set of fragments that are manageable in the underlying database. In this paper we proposed combined hill climbing and genetic algorithm in order to enhance fragmentation selection for horizontal partitioning approach. Our experimental results show that our method can provide a significantly better solution than existing fragmentation selection techniques in terms of minimization of query processing time.
展开▼
