网格环境下基于QoS需求的关联任务调度算法  

QoS-based scheduling heuristics for dependent tasks in grid computing

作　　者：姜黎立[1] 蒋昌俊[1] 袁禄来[1] 

机构地区：[1]同济大学计算机科学与工程系,上海201804

出　　处：《计算机应用研究》2008年第2期350-354,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(60173026)

摘　　要：首先描述QoS调度问题,建立QoS需求模型;然后通过分析任务的依赖性,提出时间花费、资源价格和可靠性三种QoS参数的映射机制;最后针对网格环境的新特征,提出一种以优化用户效用为目标,基于QoS的关联任务调度算法(QBDTS_UO)。仿真实验结果表明,该算法能以较小的时间花费为代价,有效满足用户的QoS需求,并能大大提高网格资源的使用率。This paper firstly described QoS scheduling problem and established a QoS demand model. Then proposed a QoS mapping mechanism for three QoS parameters which included completion time, execution cost and reliability in basis of analy. zing the relationship among tasks, Finally, proposed a computationally efficient static scheduling heuristics which was called QoS-based dependant task scheduling for utility optimization （QBDTS_UO）. Simulation results show that QBDTS_UO is capable of meeting diverse QoS requirements for users, while can greatly increase the usage of grid resources.

关 键 词：网格计算 任务调度 服务质量 关联任务 

分 类 号：TP338[自动化与计算机技术—计算机系统结构] TP301.6[自动化与计算机技术—计算机科学与技术]