网格计算环境下基于任务池的任务调度方法

任务调度是网格计算的关键技术之一。对网格计算任务调度特点及典型的任务调度策略进行了分析,提出了一种基于任务池模型的分级调度方法,该方法通过在网格系统中建立相对稳定的计算资源视图,在不同规模、级别上实现局部自治系统之间的协同工作,通过建立并维护任务池和资源池,对网格环境中动态变化的资源和任务进行管理,来保持系统资源之间的共享关系和高度可控性。实验结果表明,本调度方法具有较高的执行效率和加速比。

A task scheduling way based on the taskpool in grid computing environments

Task scheduling is one of the critical technologies of Grid Computing. Task scheduling traits and typical task scheduling strategies of grid computing are analyzed. A different level scheduling way based on the mission pool model is proposed. By creating relatively stable calculation resources view in grid system, this way can realize cooperative work among local autonomy systems on different scales and levels; By creating and maintaining the task-pool and resource-pool, this way can manage dynamic resources and tasks in grid environment to keep share relation among system resources and high controllability. The results of an experiment indicate that the proposed scheduling way has fairly good executive efficiency and the rate of acceleration.