一种基于效用函数的网格资源分配策略

Strategy for Resource Allocation in Grid Environment Based on Utility Function

针对网格资源分配中用户需求的异构性问题,提出了一种基于效用函数优化的分配策略。该策略综合考虑用户作业执行费用和执行时间两方面的因素,利用拉格朗日方法解决网格用户效用函数的优化问题,通过二分搜索最优解产生一组优化的用户出价,根据该组出价按比例划分资源的计算能力。该分配策略可对网格资源的价格以及资源的占用时间进行优化,对动态、异构的网格环境具有较好的适应性。

To solve the problem of heterogeneity of user requirements in grid resource allocation, an allocation strategy based on utility function optimization is proposed. The strategy synthetically considers two factors： execution cost and execution time of user job. The Lagrangian method is used to solve the optimal problem of user utility function. To allocate resource capacity efficiently, the allocation strategy uses a bisection search to generate a set of optimal user bids. The results show that the proposed strategy can optimize the grid resource price and grid resource processing time. The conclusion indicates that the strategy can satisfy the heterogeneity and dynamic nature of gird environment.