摘要
网格资源调度策略直接影响网格的性能,因此网格资源调度算法成为网格技术研究中的一个热点.本文针对网格的异构性,以资源调度的负载均衡、有效资源利用率的性能提高以及最小化任务集的时间跨度(Makespan)为目标,采用启发式算法的思想,通过引入预测机制推断任务集合中网格任务大小的分布情况,从而分析出任务集合的特性,根据任务集特性选择合适的资源调度策略,再通过二次分配的思想使网格系统中各个资源快速达到负载平衡和最小化任务集的时间跨度,提出一种基于标准差及二次分配的网格资源调度算法,最后采用Braun等人提出的仿真模型基准验证了算法的有效性.
Grid resource scheduling policies directly affect the performance of the grid, so the grid resource scheduling algorithm for grid research has become a hot spot. This paper based on heterogeneous grid to resource scheduling of load balancing,effective re- source utilization performance improvement and minimize the Makespan as the goal, the use of heuristics thinking by introducing infer- ence task prediction mechanism the distribution of tasks set in the grid size to analyze the characteristics of the task set, according to the task set characteristics to choose the right resource scheduling policy, and through the secondary distribution of ideas to make the grid system to quickly reach each resource load balancing and minimization time span set of tasks proposed grid resource scheduling algorithm based on the standard deviation and secondary distribution, and finally the use of simulation models Braun et al benchmarks demonstrate the effectiveness of the algorithm.
出处
《小型微型计算机系统》
CSCD
北大核心
2016年第2期259-263,共5页
Journal of Chinese Computer Systems
基金
河北省自然科学基金项目(F2012203087)资助
国家自然科学专项基金项目(61040023)资助
国家自然科学基金项目(61073060)资助
关键词
资源调度
标准差
二次分配
负载平衡
时间跨度
resource scheduling
standard deviation
secondary distribution
load balancing
makespan
