Scheduling efficiency of resource information aggregation in grid networks

P. Kokkinos, E. A. Varvarigos

Research output: Contribution to journalArticleResearchpeer-review

12 Citations (Scopus)

Abstract

We consider information aggregation as a method for reducing the information exchanged in a Grid network and used by the resource manager in order to make scheduling decisions. In this way, information is summarized across nodes and sensitive or detailed information can be kept private, while resources are still publicly available for use. We present a general framework for information aggregation, trying to identify issues that relate to aggregation in Grids. In this context, we describe a number of techniques, including single point and intra-domain aggregation, define appropriate grid-specific domination relations and operators for aggregating static and dynamic resource information, and discuss resource selection optimization functions. The quality of an aggregation scheme is measured both by its effects on the efficiency of the scheduler's decisions and also by the reduction it brings on the amount of resource information recorded, a tradeoff that we examine in detail. Simulation experiments demonstrate that the proposed schemes achieve significant information reduction, either in the amount of information exchanged, or in the frequency of the updates, while at the same time maintaining most of the value of the original information as expressed by a stretch factor metric we introduce.

Original languageEnglish
Pages (from-to)9-23
Number of pages15
JournalFuture Generation Computer Systems
Volume28
Issue number1
DOIs
Publication statusPublished - Jan 2012
Externally publishedYes

Keywords

  • Grid networks
  • Information aggregation
  • Scheduling
  • Stretch factor

Cite this