Query graphs, implementing trees, and freely-reorderable outerjoins

  • Authors:

  • Arnon Rosenthal;Cesar Galindo-Legaria

  • Affiliations:

  • ETH Zurich and Xerox AIT;Alken Computation Laboratory, Harvard University

  • Venue:
  • SIGMOD '90 Proceedings of the 1990 ACM SIGMOD international conference on Management of data
  • Year:
  • 1990

Quantified Score

Hi-index 0.00

Visualization

Abstract

We determine when a join/outerjoin query can be expressed unambiguously as a query graph, without an explicit specification of the order of evaluation. To do so, we first characterize the set of expression trees that implement a given join/outerjoin query graph, and investigate the existence of transformations among the various trees. Our main theorem is that a join/outerjoin query is freely reorderable if the query graph derived from it falls within a particular class, every tree that “implements” such a graph evaluates to the same result.The result has applications to language design and query optimization. Languages that generate queries within such a class do not require the user to indicate priority among join operations, and hence may present a simplified syntax. And it is unnecessary to add extensive analyses to a conventional query optimizer in order to generate legal reorderings for a freely-reorderable language.