Abstract
By the rapid growth of data, distributed computing and grid computing has been attracted so much attention. Grid computing is a distributed parallel processing system that resources have different architectures in it. Large number of computations that are done in a certain time in the grid can’t be done by the best supercomputers. The load balancing algorithm is used to increase the efficiency of the grid. The purpose of load balancing algorithms is to use all the resources which are available in the grid. The management of the resources in grid as a process of diagnosis and resource allocation schedules and monitors the resources. Scheduling process directs the tasks to the appropriate resources, to ensure maximum use of them. In this paper for task allocating a probability scheduling algorithm is presented for the purpose of load balancing. In this algorithm Workclass, cost, deadline and herd are considered. Probability load balancing algorithm based on the failure of resources, choses a source that had a better performance in the past, and assigns the task to it. Resources are Rewarded or punished based on their performance. In the load balancing algorithms the tasks migrate from heavy load resources to the light load ones if the tasks are imbalanced.
