Min-min Algorithm. Grid Computing To Improve

Grid computing is an important network technology in world in recent years; it will help people share resources in Internet better; task scheduling is one important part of it. Grid is dynamic, distributed, heterogeneous and diverse, which makes the task scheduling in Grid computing is more complex. This paper studies the task scheduling in grid computing in this respect.The key of task scheduling is scheduling algorithm; as a classical heuristic algorithm, Min-Min is used as the standard of performance evaluation for the other algorithms. Therefore this paper selects Min-Min algorithm to study the task scheduling in Grid computing. The contributions of this paper are as follows:1. This paper has a detailed analysis and research of Min-Min algorithm, which is firstly executing the task with minimum completion time; it's simple and has less execution time. But this strategy will cause the longer tasks waiting too long or even "starve to death", because it always scheduled shorter task earlier.2. In this respect, this paper proposes DPri-Min-Min algorithm, which introduces the concept of task dynamic priority, its initial value is assigned by user and is increased as the waiting time increase. DPri-Min-Min combines the dynamic priority and Min-Min algorithm. In the beginning of every scheduling cycle in DPri-Min-Min, all the tasks'minimum completion time and priority are calculated, and then normalized and weighted summation; at last, they are scheduled according to the weighted sum.3. In this paper, Min-Min and DPri-Min-Min algorithm are simulated in the grid simulator, GridSim. It shows that DPri-Min-Min algorithm has a significantly improved in long task scheduling, and a certain improved in the performance of Makespan and load balance; it proves that DPri-Min-Min algorithm's superiority in long task scheduling and adaptability to Grid.