网格任务调度算法研究与仿真.pdfVIP

ABSTRACT This paper introduced the concept, category, system structure, simulation toolkit of Grid ，then discussed the goals, principles and basic strategies of task scheduling. This paper analyzed familiar task scheduling algorithms of Grid, emphatically analyzed Min-Min, Max-Min, Max-Int algorithms from the angle of the performance of Grid task scheduling algorithms. In allusion to the shortage, which current task scheduling is not the best in Max-Int algorithms, the paper improved Max-Int algorithm. This algorithm arranges the task in groups, and considers the problem of next group task when it executes current group tasks，furthermore，in this course, finds the maximal loaded machine and the minimal loaded machine, then transfers or exchanges these tasks on the two machines that make the tasks balance. According to iterative operation, the algorithm makes every machine’s tasks balance in Grid system, and makes the future task scheduler run to optimization. At last, a lot of simulative experiments were made by use of GridSim-the simulation toolkit of Grid task scheduling ， and combining with the JCreator IDE 。The results of experimentation show that the improved Max-Int algorithms has good performance in Grid system compared with the algorithms of Min-Min, Max-Min and Max-Int, it’s performance has the trend of stepping up, and then resolves the shortage of Max-Int when make every task scheduling optimization every time and make future scheduling run to optimization. Keyword: Grid Computing; Resource Scheduling; GridSim; Max-Int II 目