基于资源融合的网格任务调度模型与算法

Model and Algorithms of Grid Task Scheduling Based on Resource Fusion

讨论了网格资源计算能力和网络通信速度异构的树型计算网格环境下独立任务的调度问题,为单层树型网格计算环境下的任务调度问题建立了整数线性规划模型,并采用资源融合方法将多层树型网格计算环境下的任务调度问题转化为单层树型网格计算环境下的任务调度问题,建立了相应的数学模型.基于所提出的模型构造了两个任务分配启发式算法OPCH和OPBH.实验结果表明,在异构的树型计算网格环境下实现独立任务调度时,文中提出的算法比FCFS、Min-min算法具有更好的性能.

In this paper, the scheduling of independent tasks on a tree-based grid-computing platform where resources have different speeds of computation and communication is discussed, and an integer linear programming model for the task scheduling on a fiat tree-based grid-computing platform is presented. Then, by adopting the resource fusion approach, the task scheduling on a multi-level tree-based grid-computing platform is converted into the one on a fiat tree-based grid-computing platform, and the corresponding mathematical model is proposed. Based on the proposed model, two heuristic algorithms for task allocation respectively named OPCH and OPBH are finally presented. Experimental results show that the two presented algorithms are of better performance than the First- Come-First-Served （FCFS） and the Min-min algorithms.