Optimization of Relational Expressions Using A Logical Analogon
by G. A. Blaauw, A. J. W. Duijvestijn, R. A. M. Hartmann
An expression applying to a relational database is optimized by mapping the expression upon set expressions which, in turn, are transformed into logical expressions. These logical expressions then are optimized, taking into account the constraints that are inherent in relational expressions and the costs of those expressions. Subsequently a reverse transformation to relational expressions is applied. The method is developed for the traditional relational operators and is applicable to a variety of cost criteria. Common subexpressions as well as redundant expressions are optimized. A new relational operation "split" is proposed that may be used effectively in an optimized expression. Results obtained with a model for the optimization method are presented.