论文部分内容阅读
分布式仿真网格是完善分布式仿真系统有效途径之一,而资源任务调度问题是提高分布式仿真网格系统效率的基础。通过对经典Min-Min和Max-Min调度算法研究学习,发现现有算法存在负载不均衡问题。针对现存问题,在考虑到任务本身优先级要求、任务大小和机器运行速度运行效率等因素的情况下,提出机器任务匹配度函数Matching-rate[i][j],将任务的预期完成时间与任务优先级进行匹配;并定义了系统负载均衡参数LBP,然后根据机器任务匹配度函数和系统负载均衡参数实现动态地调度Min-Min算法和Max-Min算法。最后通过实验,验证了此算法在总执行时间、总执行费用和机器利用率指标中有了一定改进,提高了调度性能,且达到了实现负载均衡的目的。
Distributed simulation grid is one of the effective ways to improve the distributed simulation system. Resource task scheduling problem is the basis of improving the efficiency of distributed simulation grid system. By studying and studying the classical Min-Min and Max-Min scheduling algorithms, it is found that the existing algorithms have the problems of unbalanced load. Considering the priority of the task itself, the size of the task, and the operating efficiency of the machine running speed, aiming at the existing problems, the machine task matching function Matching-rate [i] [j] is proposed. The expected completion time of the task Task priority. The system load balancing parameter LBP is defined, and then the Min-Min algorithm and the Max-Min algorithm are dynamically scheduled according to the machine task matching degree function and the system load balancing parameter. Finally, experiments show that this algorithm has some improvements in the total execution time, total execution cost and machine utilization index, which improves the scheduling performance and achieves the purpose of load balancing.