Query Optimization in Distributed Relational Databases

  • Authors:

  • Celso C. Ribeiro;Cláudio D. Ribeiro;Rosana S. G. Lanzelotte

  • Affiliations:

  • Pontifícia Universidade Católica do Rio de Janeiro, Departamento de Informática, Rua Marquês de São Vicente 225, Rio de Janeiro 22453, Brazil. E-mail: celso@inf.puc-rio.b ...;Pontifícia Universidade Católica do Rio de Janeiro, Departamento de Informática, Rua Marquês de São Vicente 225, Rio de Janeiro 22453, Brazil. E-mail: celso@inf.puc-rio.b ...;Pontifícia Universidade Católica do Rio de Janeiro, Departamento de Informática, Rua Marquês de São Vicente 225, Rio de Janeiro 22453, Brazil. E-mail: celso@inf.puc-rio.b ...

  • Venue:
  • Journal of Heuristics
  • Year:
  • 1997

Quantified Score

Hi-index 0.00

Visualization

Abstract

The query optimizer is the DBMS (data base management system)component whose task is to find an optimal execution planfor a given input query. Typically, optimization is performed usingdynamic programming. However, in distributed execution environments,this approach becomes intractable, due to the increase in the searchspace incurred by distribution. We propose the use of the tabu searchmetaheuristic for distributed query optimization. A hashing-baseddata structure is used to keep track of the search memory,simplifying significantly the implementation of tabu search. Tovalidate this proposal, we implemented the tabu search strategy inthe scope of an existing optimizer, which runs several searchstrategies. We focus our attention on the more difficult problems interms of the query execution space, in which the solution spaceincludes bushy execution plans and Cartesian products, which are notdealt with very often in the literature. Using a real-lifeapplication, we show the effectiveness of tabu search when comparedto other strategies.