摘要
任务调度是网格应用系统获得高性能的关键.网格计算中一个大型的应用程序往往被分解为具有依赖关系的多个任务.在资源个体差异较大、广域互连的网格环境下任务间的依赖关系对传统的调度策略提出了新的挑战.任务调度的主要工作是为任务分配资源以及确定任务的执行次序,将依赖任务的可能的资源分配方案表示为任务-资源分配图(T-RAG),在该图的基础上提出了基于T-RAG优化选取的依赖任务调度模型,将依赖任务调度问题转化为图的优化选取问题,解析最优任务-资源分配图可以同时确定资源分配方案和任务的执行次序即为最优调度方案.最后,实现了基于该模型的任务调度算法,该算法与ILHA算法的对比分析表明,在资源差异较大及任务间存在大量数据传输的情况下所提出的算法更优.
Efficient task scheduling is critical for grid application to achieve high performance.In grid computing,an application is decomposed into a set of dependent tasks.In the grid environment where resources have different capability and resources are interconnected over the world,the dependence among tasks affects the scheduling strategy greatly.The general task scheduling problem includes the problem of assigning the tasks of an application to suitable resource and the problem of ordering task executions on each resource.In this paper a task-resource assignment graph（T-RAG） is used to represent a potential resource assignment plan.And a model of task scheduling based on optimization selection of T-RAG is proposed,which maps the dependent task scheduling problem into a graph optimization problem.As a result of this optimization selection,the optimal graph is obtained and such optimal graph is the optimal scheduling plan which determines the resource assignment plan and the execution order of tasks.Finally,the task scheduling algorithm based on the proposed scheduling model is implemented.Compared with the ILHA algorithm in the simulation environment,the proposed algorithm shows better performance in the situation of a large body of data transported among tasks and significant differences in resources capability and network bandwidth.
出处
《计算机研究与发展》
EI
CSCD
北大核心
2007年第10期1741-1750,共10页
Journal of Computer Research and Development
基金
国家"十五"重点科技攻关基金项目(2004BA721A05)
关键词
网格计算
任务调度
依赖任务
遗传算法
任务-资源分配图
grid computing
task scheduling
dependent task
genetic algorithm
task-resource assignment graph
作者简介
陈廷伟,1976年生,博士,主要研究方向为网格计算和服务计算.（t.w.chen@163.com）
张斌,1964年生,教授,博士生导师,中国计算机学会高级会员,主要研究方向为服务计算、信息集成.
郝宪文,1975年生,博士研究生,主要研究方向为网格计算和服务计算.
