GangSim: A Simulator for Grid Scheduling Studies with support for uSLAs

   Large distributed Grid systems pose new challenges in job scheduling due to complex workload and system characteristics. Due to the numerous parameters that must be considered and the complex interactions that can occur between different resource allocation policies, analytical modeling of system behaviors appears impractical.
   GangSim is a tool developed for Grid scheduling studies, capable of supporting studies for controlled resource sharing based on uSLAs. The starting point of this work was an exploration of distributed monitoring systems conducted in the context of the GriPhyN/iVDGL projects. I started by developing the VO-Ganglia Monitoring Toolkit to gather resource characteristics and usage limits to meet the requirement needs for uSLA-based resource sharing. The result was an enhancement of the Ganglia Monitoring Toolkit. From VO-Ganglia was relatively easy to replace "real sites" with "simulated sites' and thus to enable the evaluation of a wider range of possibilities for Grid scheduling than it is possible in a real system. The new neame, GangSim, reflects both the origins of the implementation and the fact that it can be used to simulate "gangs" of consumers and resources.
   The simulated environment is presented in a schematic way in the following figure.

  • June 2003: VO-Ganglia is available for download here
  • June 2005: A first version of GangSim is available for download here

  • Project Participants:
    • to Rober Gardner, and Ian Foster for suggestions

    Some already published results (or using Ganglia) are: A qualitative comparison with other Grid simulators can be found here.

    Last Updated: 06/01/2006

    Free Hit Counter
    Free Hit Counter