Selection and advanced reservation of backup resources for high availability service in computational grid

Abstract

Resource redundancy is a primary way to improve availability for applications in computational grid. How to select backup resources for an application during the resource allocation phase and how to make advanced reservation for backup resources are challenging issues for grid environment to provide high availability service. In this paper, we proposed a backup resources selection algorithm base on resources clustering, then devised several policies for advanced reservation of backup resources. With this algorithm and these policies, the backup resource management module in grid middleware can provide resource backup service more efficiently and cost-effectively. These algorithm and policies can be implemented on the GARA system, make the QoS architecture more powerful and practical for computational grid. © Springer-Verlag 2004.