Effective scheduling algorithm to reduce total completion time and promote resource utilization with load balancing in a grid computing environment is required. Scheduling tasks on heterogeneous machines distributed over a grid system proves to be an NP complete problem. Many algorithms have been developed to counter this problem by researchers. However, it is obvious that, task selection is a key challenge to these heuristics. For this reason, a substantial enhancement in the computational efficacy of the algorithm might be welcome. In this paper, a new batch mode scheduling algorithm (MinExt) is proposed. The intent is to reduce the total completion time (makespan), utilization of idle resources and load balance. To achieve this, the proposed algorithm made an initial task queue, we collects the Average Completion Time (Act) of all tasks, then for all tasks greater than Act is scheduled first and follow by the set of tasks less than or equal to the Act. Our simulation results indicate that the algorithm minimizes total completion time and utilizes the idle resources effectively with load balancing in comparing to other algorithms.
展开▼
