Automatic relationship discovery in self-managing database systems

Abstract

Various aspects of CORDS, an algorithm that automatically discovers correlations and soft functional dependencies (FD) between pairs of columns, are discussed. CORDS automatically ranks the discovered soft FDs and correlations by an appropriate measure of strength. CORD applies a flexible set of heuristic pruning rules to reduce the search space and avoid discovery of spurious relationships. It is observed that in developing CORDS, algorithmic simplicity and judicious use of sampling can lead to efficient and highly scalable self-management algorithms that are suitable for immediate incorporation into commercial systems.