Abstract | ||
---|---|---|
Many researchers focus on resource intensive tasks which have to be run continuously over long periods. A Grid may offer resources for these tasks, but they are contested by multiple client agents. Hence, a Grid might be unwilling to allocate its resources for long terms, leading to tasks' interruptions. This issue becomes more substantial when tasks are data inter-dependent, where one interrupted task may cause an interruption of a bundle of other tasks. In this paper, we discuss a new resource re-allocation strategy for a client, in which resources are re-allocated between the client tasks in order to avoid prolonged interruptions. Those re-allocations are decided by a client agent, but they should be agreed with a Grid and can be performed only by a Grid. Our strategy has been tested within different Grid environments and noticeably improves client utilities in almost all cases. |
Year | DOI | Venue |
---|---|---|
2016 | 10.1007/978-3-319-59294-7_16 | Lecture Notes in Artificial Intelligence |
Keywords | Field | DocType |
Continuous inter-dependent tasks,Resource re-allocation,Client's decision-making mechanism | Computer science,Long terms,Grid,Bundle,Distributed computing | Conference |
Volume | ISSN | Citations |
10207 | 0302-9743 | 0 |
PageRank | References | Authors |
0.34 | 7 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Valeriia Haberland | 1 | 10 | 3.91 |
Simon Miles | 2 | 1599 | 109.29 |
Michael Luck | 3 | 3440 | 275.97 |