E. Xia and I. Jurisica (Canada)
Grid Scheduling, Grid Configuration, Application Performance.
In the grid environment, applications make use of shared grid resources to improve performance. The target func tion usually depends on many parameters, e.g., the schedul ing strategies, the configurations of machines and links, the workloads in a grid, the degree of data replication, etc. In this paper, we examine how those parameters may affect performance. We choose an application's overall re sponse time as an object function and focus on dynam ically scheduling independent tasks. We define the job, scheduler and cost models and conduct experiments using simulations. Understanding influence of each parameter is not only crucial for an application to achieve good perfor mance, but may also help to develop effective scheduling heuristics and design high quality grids.
Important Links:
Go Back
