一种改进的集群系统的任务调度算法  

Algorithm of improved cluster system task scheduling

作　　者：张亭[1] 

机构地区：[1]天津美术学院设计与新媒体艺术实验教学中心,天津300342

出　　处：《实验室科学》2013年第3期82-84,88,共4页Laboratory Science

基　　金：天津美术学院科研项目(项目编号:201143)

摘　　要：任务调度算法是提高集群系统负载均衡能力的有效手段。为了提高系统利用效率,除了每个任务分配优先级外,还提出基于动态分配任务抢占阈值的LSF(Least Slack First最小空闲时间优先算法)改进算法,并将该设计方法应用到渲染集群系统中,从而有效地减少了因任务抢占引起的系统开销和提高了渲染集群系统资源利用率。Task scheduling algorithm is an effective means to improve the load balancing capabilities of the cluster system . Use efficiency in order to improve the system, in addition to each task assigned priority, this paper also proposed preemption threshold based on dynamic allocation of tasks LSF improved algorithm, and the design method applied to the rendering cluster system, which effectively reduces the task preemption caused overhead and improve the rendering cluster system resource utilization.

关 键 词：负载均衡 调度算法 动态抢占阈值 

分 类 号：TP368[自动化与计算机技术—计算机系统结构]