Resource Management in Grids: Overview and a discussion of a possible approach for an Agent-Based Middleware

Mehrdad Senobari (1) , Michal Drozdowicz (2) , Maria Ganzha (2) , Marcin Paprzycki (2) , Richard Olejnik (3, 4) , Ivan Lirkov (2, 5) , Pavel Telegin (6) , Nasrollah Moghadam Charkari (1)

1 Tarbiat Modares University [Tehran]
2 IBS PAN - Systems Research Institute [Warsaw]
3 DART - Contributions of the Data parallelism to real time
4 LIFL - DART/Émeraude
5 IPP BAS - Institute for Parallel Processing
6 SuperComputing Center of Russian Academy of Sciences

Resource management and job scheduling are important research issues in computational grids. When software agents are used as resource managers and brokers in the Grid a number of additional issues and possible approaches materialize. The aim of this chapter is twofold. First, we discuss traditional job scheduling in grids, and when agents are utilized as grid middleware. Second, we use this as a context for discussion of how job scheduling can be done in the agent-based system under development.

Mots clés: grid computing; agent-based resource management; resource management; job scheduling; metaschedulers

Domaines: Calcul parallèle, distribué et partagé [cs.DC]; Système multi-agents [cs.MA]

Mehrdad Senobari, Michal Drozdowicz, Maria Ganzha, Marcin Paprzycki, Richard Olejnik, et al.. Resource Management in Grids: Overview and a discussion of a possible approach for an Agent-Based Middleware. Saxe-Coburg Publications, Stirlingshire, UK. PARALLEL, DISTRIBUTED AND GRID COMPUTING FOR ENGINEERING, Saxe-Coburg Publications, Stirlingshire, UK, pp 141-164, 2009, Computational Science, Engineering & Technology Series, ISSN 1759-3158. ⟨10.4203/csets.21.8⟩. ⟨hal-00834409⟩