Fair Grid Scheduling

Abstract : Computing grids allow sharing of distributed resources such as CPU or storage to the service of many scientific communities. Users submit irregular bursts of jobs. This irregula rusage of the grid allow an efficient sharing of resources because unused resources can be collected this way. Jobs are mono-processors and arrive on-line to clusters. Jobs on a cluster are independents, they could be run in any order. Pre-emption is not used because of memory and communication costs.
docType_s :
Poster communications
3rd EGEE User Forum, Feb 2008, Clermont-Ferrand


http://hal.in2p3.fr/in2p3-00336878
Contributor : Jeanine Pellet <>
Submitted on : Wednesday, November 5, 2008 - 2:00:06 PM
Last modification on : Wednesday, November 5, 2008 - 3:00:04 PM

Identifiers

  • HAL Id : in2p3-00336878, version 1

Citation

E. Medernach. Fair Grid Scheduling. 3rd EGEE User Forum, Feb 2008, Clermont-Ferrand. <in2p3-00336878>

Export

Share