摘要
网格环境中资源的负载预测是实现资源优化分配的关键任务之一,而网格资源的动态性和异构性使得准确判断资源的负载状态十分困难.针对已有的分配策略对资源负载评估的不足,提出了一种基于序贯博弈的优化用户时间的网格资源分配策略.该策略将正比例资源共享的网格环境中多用户竞争同一计算资源的问题形式化为一个多人序贯博弈,通过寻求该序贯博弈中各个阶段博弈的纳什均衡解来预测资源负载;然后利用此负载信息生成所有用户的最优出价组合和资源的优化价格;最后根据各用户出价,按比例分配资源的计算能力.通过对网格模拟器GridSim的实验研究,结果表明,该策略能够得到合理的用户出价,降低资源占用时间,从而弥补了Bredin提出的优化策略中未考虑资源未来负载变化的缺陷,实现了资源的优化分配.其结论说明运用序贯博弈方法预测资源负载是可行的,且能更好地适应网格环境下异构资源的动态性.
In grid environment, resource load prediction is one of the most important problems in resource allocation optimization. But load status is difficult to estimate accurately due to the dynamic nature and heterogeneity of grid resource. In response to this issue, a resource allocation strategy that uses sequential game method to predict resource load for time optimization in a proportional resource sharing environment is proposed. The problem of multiple users bidding to compete for a common computational resource is formulated as a multi-player dynamic game. Through finding the Nash equilibrium solution of the multi-player dynamic game, resource load is predicted. Using this load information, a set of user optimal bids is produced to partition resource capacity according to proportional sharing mechanism. The experiments are performed based on the GridSim toolkits and the results show that the proposed strategy could generate reasonable user bids, reduce resource processing time, hence overcome the deficiency of Bredin＇s strategy, which is not concerned with resource load variation. The conclusion indicates that employing sequential game method for load prediction is feasible in grid resource allocation and adapts better to the dynamic nature of heterogeneous resource in grid environment.
出处
《软件学报》
EI
CSCD
北大核心
2006年第11期2373-2383,共11页
Journal of Software
基金
国家自然科学基金No.50479055~~
