基于移动Agent和混合蛙跳算法的网格计算资源管理

Grid Resource Management Based on Mobile Agent and Shuffle Frog Leaping Algorithm[HS)]

摘要:

为了克服传统网格系统采用Globus集中式模型或P2P网络分布式模型具有的无法兼顾全局控制性和扩展性的问题,提出了一种基于移动Agent和混合蛙跳算法的网格资源管理模型；首先,引入移动Agent将资源管理模型划分为应用管理模块、资源管理模块和资源表示模块,然后通过在经典的混合蛙跳算法加入差分扰动设计了一种改进的混合蛙跳算法,最后,在资源管理模块中的资源分配Agent中运行改进的混合蛙跳算法实现用户任务到资源的调度；采用GridSim工具进行仿真,实验结果表明,文中的资源管理模型具有较强的全局控制能力和扩展性能,资源利用率高达95.65%,较其它方法具有较小的时间跨度和较高的资源利用率,是一种适用于网格环境的有效资源管理模型。

Abstract:

In order to conquer the traditional grid system using the Globus concentration model or P2P model can not give consideration of both global control performance and extension ability, a grid resource management model based on mobile agent and shuffled frog leaping algorithm was proposed. Firstly, the resource model was divided to application management model, resource management model and resource representation model, then an improved algorism based on the classic shuffled frog leaping algorithm was proposed by adding the difference disturbance. Finally, the resource allocation Agent in the resource management model was operating the improved shuffled frog leaping algorithm to realize the task scheduling to the resource. The GridSim tool was used to simulate the method in this paper, and the result shows our method can realize the resource scheduling and the resource usage rate is as high as 95.65%, and compared with other methods, it has the least makespan and highest resource usage ratio, so it is an effective model in the grid environment.