Abstract:
Computational grids have the potential computing power for solving large-scale scientific computing applications. To improve the global throughput of these applications, workload has to be evenly distributed among the available computational resources in the grid environment. So load balancing becomes one of the critical issues that must be considered in managing a grid computing environment. Hence we need to develop a robust and effective load balancing application which can adopt changes dynamically, because due to the distributed and heterogeneous nature of the resources in grid availability of grid resources is dynamic.
In this dissertation a decentralized load balancing algorithm for computational grid is proposed. It efficiently handles the load in grid environments with considering several other issues that are imperative to Grid environments such as handling resource heterogeneity, communication latency, and job migration from one site to other. The algorithm uses the system parameters such as the estimated completion time of task, CPU processing power, load on the resource, and predicted failure time of the resource and balance the load by migrating jobs from over loaded resources to underloaded or idle resources by taking into account the job transfer cost, resource heterogeneity, and network heterogeneity . The performance of the proposed algorithm is evaluated by using several influencing parameters such as the number of jobs, job size, data transfer rate, and migration limit. The experimental results shows that the proposed algorithm is efficient in minimizing the response time ,total execution time with maximum resource utilization and minimum communication over head
