一种基于效用函数优化的计算网格资源分配策略

A Computational Grid Resource Allocation Policy Based On Utility Function Optimization

提出了一种基于效用函数优化的网格资源分配策略.试图利用效用函数和竞标函数对网格资源进行合理分配,采用拉格朗日方法对网格任务A gen t效用函数进行优化,从而可使网格任务A gen t在能估计资源节点拥塞度,并能完成其所有任务的情况下,产生一个合理的费用预算.对网格任务A gen t的竞标函数的特征进行了研究分析,研究结果表明如果网络状态不变,网格任务A gen t不能通过改变投标来获取效益.这种恒定性在任务A gen t不知晓竞标结果的情况下,可使它无需做徒劳的重投标.

This paper presents a computational grid resource allocation policy based on utility function optimization, which will apply utility functions and bid functions to allocate grid resources rationally. It can apply Lagrange methods to optimize grid agent＇s utility function, the optimization policy is to enable grid agent to estimate the congestion state of grid resource nodes, complete its tasks and produce rational budget. The characteristics of grid task agent＇s bid function are also analyzed, and the results show that if the state of network doesn＇t change, grid task agent cannot get benefits through changing bid. The invariance makes the grid task agent not to fruitlessly rebid, providing the agent doesn＇t know the results of the bidding.