On the Distributed Database Union Operation

Abstract

In a distributed relational database, an operation that produces the union of fragments stored at different sites at a specified site is called a distributed union operation. In implementing the distributed union operation, operations other than the union operation, such as intersection and difference, may be used to facilitate the union. Optimization of the distributed union operation when the fragments are not disjoint requires the ability to predict the size of the result of set operations. A minimal basis of the result of multiple non-disjoint horizontal fragments of a relation is derived. This is used to develop a model for the optimization of the distributed union operation. As the problem is NP Hard, heuristic and lower bounding procedures are developed for a special case.Request access from your librarian to read this article's full text.