Charalampos Manifavas, Nikos Papadakis, Antonis Misargopoulos, and Maria Psaraki
Join queries, query plan, grid resource allocation, distributedarchitectures
A Grid is a collection of computing resources that share and perform tasks. Performance optimization can be achieved by exploiting the Grid power utilities, i.e., multi-processing, multi-programming and abstract task execution. This paper explores the management of distributed relational databases over Grid infrastructures for the execution of ϑ−join queries. Our goal is to present an efficient resource allocation model concerning both computation and communication cost to be used for highly performing dynamic scheduling. We propose the Query Plan Graph Constructor algorithm for the query plan graph construction and the Heuristic Query Path Selection (HQuPaS) algorithm for the query plan selection. In addition, we present refined heuristic query path selection, a refined algorithm based on HQuPaS that reduces the number of states the algorithm has to visit. Finally, we present some experimental results from a scheduling simulator we implemented to evaluate these algorithms.
Important Links:
Go Back
