Subquery allocations in distributed databases using genetic algorithms

Abstract

Minimization of query execution time is an important performance objective in distributed databases design. While total time is to be minimized for On Line Transaction Processing (OLTP) type queries, response time has to be minimized in Decision Support type queries. Thus different allocations of subqueries to sites and their execution plans are optimal based on the query type. We formulate the subquery allocation problem and provide analytical cost models for these two objective functions. Since the problem is NP-hard, we solve the problem using genetic algorithm (GA). Our results indicate query execution plans with total minimization objective are inefficient for response time objective and vice versa. The GA procedure is tested with simulation experiments using complex queries of up to 20 joins. Comparison of results with exhaustive enumeration indicates that GA produced optimal solutions in all cases in much less time.