Abstract

Hierarchical grid computing is a way to gain high computing power at low cost. Efficient resource selection in such grids is challenging due to the large number of distributed resources. In this study a hierarchical structure of schedulers is described to improve the scalability of the grid system. The grid consists of clusters and each cluster is represented by a local scheduler. Each local scheduler first attempts to schedule the job in its cluster and if this fails, communicates with the grid scheduler to perform transfer or reception of job. A resource scheduling model with Minimum Total Time to Release (MTTR) strategy is proposed which follows a hierarchical two-layered approach, First layer is called grid level and other layer is called cluster level. TTR includes the processing time of the job, waiting time in the queue, transfer of input and output data to and from the resource. In this study three steps strategy has been used to determine a resource for an arriving job. It also determines the distribution of job to the remote cluster for optimizing the performance. The main objective is to obtain a minimum TTR through effective management of available grid resources. The proposed hierarchical job distribution strategy minimizes the drawbacks associated with the centralized policy while retaining its advantages. The results demonstrate that the performance of hierarchical resource selection policy is significantly greater than that of a centralized policy.

Paper Details