Gang调度在云计算任务迁移和饥饿处理中的性能和成本评估  

Gang scheduling performance and cost assessment task migration and starvation in cloud computing

作　　者：黄坤[1] 吴俊[1] 

机构地区：[1]义乌工商职业技术学院,浙江义乌322000

出　　处：《华中师范大学学报（自然科学版）》2013年第5期621-625,631,共6页Journal of Central China Normal University：Natural Sciences

基　　金：浙江省科技创新人才计划项目(2010R30044)

摘　　要：基于亚马逊的弹性计算云(EC2)的架构,对云计算模型中群调度的应用进行研究.研究中整合了作业迁移和作业饥饿处理机制,并同时考虑其性能与成本.在任意时刻可用的虚拟机的数量处于动态变化中,其规模取决于正在被服务的作业的需求.通过仿真研究某些模型,以分析具有作业迁移和饥饿处理机制的群调度的性能与总体成本.结果充分显示该调度策略可被有效地部署在云端,云平台可用于高性能计算或高性能的企业应用中.Based on Amazon＇s Elastic Computing Cloud（EC2） architecture,group scheduling has been researched in the cloud computing model.The integrated job migration and jobs hunger processing mechanism has taken into account the performance and cost.At any time,the number of available virtual machines in a dynamic change and its size depends on the needs of the job being serviced.In this paper,mentioned model has been analyzed and studied the performance and the overall cost of the group scheduling with job migration and hunger handing mechanism.Experimental results showed that the scheduling policy could be effectively deployed in the cloud and the cloud platform could be used for high performance enterprise applications.

关 键 词：云计算 Gang调度 任务迁移 饥饿处理 

分 类 号：TP393.02[自动化与计算机技术—计算机应用技术]