Applying rules for partitioned parallelism in OODBMS within an optimizer generator framework

Carlo Giovano PiresJavam C. Machado

This work presents a rule-based approach for declarative query optimizer generation considering parallel execution in object-oriented databases. The main goal of this work is to provide a framework that can capture relevant aspects of parallel query optimization in a declarative way, combining procedural techniques with the advantages of rule processing. One of those techniques was used for determining repartitioning and selecting the algorithms to evaluate operator in a query tree considering the trade-o?s between processing costs and repartitioning costs. This technique was proposed as a procedural algorithm and it was adapted to be used as rules in the context of object-oriented database optimizers. Another algorithm, used for query operator reordering in object-oriented databases, was adapted in order to consider repartitioning cost and was added to the framework. Finally, a new module for processing rules for parallelism extraction was added to the framework, providing a better support for inter-operator parallelism optimization techniques.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: