An optimization of resource replication access in grid cache

Authors:
Fangai Liu;Fenglong Song
Affiliations:
Department of Computer Science, ShanDong Normal University, JiNan, China;The Institute of Computing Technology, China Academy of Science, Beijing, China
Venue:
GPC'08 Proceedings of the 3rd international conference on Advances in grid and pervasive computing
Year:
2008

Citing 4
Cited 1

Evaluation of an Economy-Based File Replication Strategy for a Data Grid

CCGRID '03 Proceedings of the 3st International Symposium on Cluster Computing and the Grid
Accurate Modeling of Cache Replacement Policies in a Data Grid

MSS '03 Proceedings of the 20 th IEEE/11 th NASA Goddard Conference on Mass Storage Systems and Technologies (MSS'03)
Towards an Economy-Based Optimisation of File Access and Replication on a Data Grid

CCGRID '02 Proceedings of the 2nd IEEE/ACM International Symposium on Cluster Computing and the Grid
The Anatomy of the Grid: Enabling Scalable Virtual Organizations

International Journal of High Performance Computing Applications

End-To-End Cache System for Grid Computing: Design and Efficiency Analysis of a High-Throughput Bioinformatic Docking Application

International Journal of High Performance Computing Applications

Quantified Score

Hi-index	0.00

Visualization

Abstract

In a grid system, resources are located in different nodes and a grid member can obtain its resources from different positions. But the disproportion between the resource distributions and its property of heterogeneity cause communication bottlenecks, which can decrease the grid performance. To solve this problem, cache techniques have been proposed. However, in a grid, how to access the resources in an economic way and how to decide whether one resource replica should be cached, are the two long standing problem. In this paper, an economic model is proposed to optimize the cache management policy. A parallel algorithm of accessing data from m resource sites is improved, where we propose a strataties on how to select the m resource sites to access the data in parallel. By this means, the single site failure problem can be solved. The test results show that the strategies we propose are efficient.