Optimization of Queries Including ADT Functions
Kenichi Yajima, Hiroyuki Kitagawa, Kazunori Yamaguchi, Nobuo Ohbo, Yuzuru Fujiwara · Database Systems for Advanced Applications · 1991
Inclusion 01 ADTs (AbsWacr Dula Types) has been studied by many researchers as a promising feature to make the relational database meet requirements from non-traditional advanced applications. In databases which support ADTs, the execution cost of a selection involving ADT functions (shortly, ADT selecdun) may be equal to or possibly more than that of a join, since the evaluation of a computationally and expensive complex function is often required in the test of the selection condition. In such environments, ADT selections become dominant cost factors in query optimization. The conventional query optimization heuristics sometimes does not work well in such databases. since it focuses on joins as dominant cost factors. In this paper, we propose an optimization method that takes ADT selections as well as @ins into consideration. Based on some simulation results, we show that our method will find query execution plans superior to those obtained by the conventional heuristics, in particular, 1) when ADT selections are considerably more costly than joins, and/or 2) when there exist more than one ADT selection in a given query. We also propose some schemes to reduce the optimization cost of our method.