A method for hierarchy processing in relational systems

摘要

Processing of hierarchies has always been a difficult task in relational database environments. Several approaches have been proposed, but they have focused their attention mostly on improving the semantic power of relational systems in order to express recursive definitions on hierarchies rather than propose feasible computing methods. In this paper we show an encoding method which allows us to cope with one-to-many hierarchical relationships in a very different way. The novel idea is not to store the key of the parent tuple but create a code for the path of all the ancestor tuples. Two major advantages are thus obtained: a considerable reduction in the retrieval of unnecessary intermediate tuples, and an enrichment of the information that is utilized by the optimizer module to choose the best access strategy.