Abstract

Over the last decade, many techniques for optimizing relational queries have been developed. However, the optimization of queries with aggregation has received little attention. This paper investigates possible improvements for aggregate queries on groups of tuples. We suggest the use of program transformation methods to systematically generate efficient iterative programs for their evaluation. Two transformation steps successively translate a program, which sorts the relation before applying the aggregate function, into an iterative program which performs the aggregation while sorting, thus using less time and space than needed for the execution of the initial program.

Citations