Evaluation of Upper Bounds and Least Nodes as Database Operations

We argue for supporting besides transitive closure two additional types of recursive queries in augmented relational database systems: the computation of upper bounds and least nodes in graphs that are persistently stored as database relations. We describe various algorithms for the evaluation of the specified query types. In essence, they vary in the kind of graph search, the search direction, the way how they prune search areas, and the way how they utilize intermediate results. Based on analytical results, we give a comparative assessment of the algorithms. The results of experiments substantiate our analytical findings.