Paper Info
Title
Heuristic-based scheduling to maximize throughput of data-intensive grid applications
Abstract
Job scheduling in data grids must consider not only computation loads at each grid node but also the distributions of data required by each job. Furthermore, recent trends in grid applications emphasize high throughput more than high performance. In this paper, we propose a centralized scheduling scheme, which uses a scheduling heuristic called Maximum Residual Resource (MRR) that targets high throughput for data grid applications. We have analyzed the performance potentials of MRR, and have developed a simulator to evaluate it with typical grid configurations. Our results show that MRR brings significant performance improvements over existing online and batch heuristics like MCT, Min–min and Max-min.
Year
DOI
Venue
2004
10.1007/978-3-540-30536-1_8
IWDC
Keywords
Field
DocType
centralized scheduling scheme,data grid,performance potential,high throughput,job scheduling,grid node,data-intensive grid application,heuristic-based scheduling,data grid application,high performance,grid application,typical grid configuration
Residual,Heuristic,Scheduling (computing),Computer science,Data grid,Heuristics,Job scheduler,Throughput,Grid,Distributed computing
Conference
Volume
ISSN
ISBN
3326
0302-9743
3-540-24076-4
Citations 
PageRank 
References 
1
0.35
17
Authors
2
Name
Order
Citations
PageRank
Souvik Ray142.09
Zhao Zhang2797.53