Performance comparisons and scheduling of load balancing strategy in Grid Computing

In a distributed Grid computing system it is desirable to achieve a uniform distribution of the workload among the machines in such a way that each machine would have the same workload. No machine should remain idle, while other machines are overloaded. The problem of load distribution mainly involves reallocating the workload among the host machines of the distribution system. Load definition is used to obtain better response time and to increase resource utilization, and thus to achieve improved performance. This can be achieved by using process migration strategy. Our focus is on dynamic load balancing on a network of workstations (Grid Computing) and to develop a distributed scheduling algorithm for load balancing which takes heterogeneity CPU, memory and disk resource into account.