The grid: blueprint for a new computing infrastructure
The grid: blueprint for a new computing infrastructure
Heuristic Algorithms for Scheduling Independent Tasks on Nonidentical Processors
Journal of the ACM (JACM)
Dynamic mapping of a class of independent tasks onto heterogeneous computing systems
Journal of Parallel and Distributed Computing - Special issue on software support for distributed computing
IEEE Transactions on Parallel and Distributed Systems
Journal of Parallel and Distributed Computing
The Impact of More Accurate Requested Runtimes on Production Job Scheduling Performance
JSSPP '02 Revised Papers from the 8th International Workshop on Job Scheduling Strategies for Parallel Processing
Segmented Min-Min: A Static Mapping Algorithm for Meta-Tasks on Heterogeneous Computing Systems
HCW '00 Proceedings of the 9th Heterogeneous Computing Workshop
Scheduling functional regression tests for IBM DB2 products
CASCON '05 Proceedings of the 2005 conference of the Centre for Advanced Studies on Collaborative research
CasSim: a top-level-simulator for grid scheduling and applications
CASCON '06 Proceedings of the 2006 conference of the Center for Advanced Studies on Collaborative research
Backfilling Using System-Generated Predictions Rather than User Runtime Estimates
IEEE Transactions on Parallel and Distributed Systems
Are user runtime estimates inherently inaccurate?
JSSPP'04 Proceedings of the 10th international conference on Job Scheduling Strategies for Parallel Processing
| Hi-index | 0.00 |
It has been shown that runtime estimation errors have a large impact on scheduler performance. In previous research, scheduling algorithms were mainly used in a homogeneous environment. In this paper, we investigate several scheduling heuristics that are commonly used in the grid environment. We systematically study how runtime relative estimation errors affect the scheduler performance in different grid scenarios by conducting experiments using simulation. We choose Dynamic-selection, Min-min, Seg-min-min, Max-min, and Sufferage as our scheduling algorithms for the experiments. Our results show interesting trends: (1) increased estimation error results in degrading performance of all tested scheduling heuristics, making them even worse than the basic "Round-Robin" approach if errors are large; however, locally, performance is sometimes better and, in some special cases, estimation errors do not affect scheduler performance; (2) unlike in general, increased estimation errors diminish the performance difference among individual heuristics; (3) there is a performance threshold, no matter how large the estimation errors are; (4) increased accuracy of runtime estimation improves performance in general.