The grid: blueprint for a new computing infrastructure
The grid: blueprint for a new computing infrastructure
DSC: Scheduling Parallel Tasks on an Unbounded Number of Processors
IEEE Transactions on Parallel and Distributed Systems
ACSW Frontiers '05 Proceedings of the 2005 Australasian workshop on Grid computing and e-research - Volume 44
Concurrency and Computation: Practice & Experience
| Hi-index | 0.00 |
The past decade has witnessed a boost in the acceptance of Grid Computing as an alternative to traditional supercomputing environments. The evolution of Computers, Networks and the Internet has boosted the growth of Grid Computing. However applications with fine grained jobs tend to be unfavorable for this environment. Further applications that possess jobs that are dependent upon each other need to be grouped together and represented in a coarse-grained manner to lower the overheads involved with fine grained jobs, and rescheduling of dependent jobs. Also, there is a need to impose ordering on resource selection based on their processing capabilities and dynamic bandwidth scenarios. Users with different priorities must be provided differentiated Network ToS to ensure high Success Rates of job completion. In this paper, we investigate a dependency based, Network Aware, dynamic job grouping scheduler, which provides Network ToS to Users based on their deadlines.