基于网格资源可用度的元任务映射算法
Meta-Task Mapping Heuristic Based on Serviceability of Grid Resource
-
摘要: 针对Min-Min启发式算法没有充分考虑网格的异构特性, 导致资源负载不均衡的问题,提出一种基于资源可用度的启发式映射算法,在元任务与资源映射过程中,先计算资源对元任务的可用度,然后将元任务划分为优先映射与次优先映射两部分,再对这两组元任务分别应用经典Min-Min算法进行映射. 采用仿真软件模拟网格异构环境,在元任务与资源一定的情况下,分别模拟这两种算法的元任务与资源的映射过程,得到这两种情况下的任务完成时间. 实验结果表明,基于资源可用度的启发式算法在异构的网格环境中任务完成时间更短.Abstract: Without taking the heterogeneous grid characteristics into account, Min-Min heuristic can lead to load imbalance. A novel mapping heuristic based on resources serviceability is proposed in this work. The mapping process is as follows. First, the serviceability of resources to meta-task is calculated and then the meta-task is divided into two groups-prior group and secondary group. Two groups of meta-task are mapped by use of classical Min-Min heuristic, respectively. Under the condition of simulated heterogeneous grid environment with fixed meta-task and resources, the experiments of using classical heuristic and proposed heuristic were conducted to obtain meta-task makespans. The results show that novel mapping algorithm based on resources serviceability has shorter makespan under heterogeneous grid environment.
下载: