摘要
分布式计算环境中可将大作业进行任务分解,对分解后的一系列短作业采用最优化调度策略以达到缩短大作业整体周转时间和系统响应时间目的。针对传统调度策略的不足及云计算中网络延迟较大的特点,拟在云计算环境虚拟层对资源进行重新分配,根据自定义当前虚拟主机(KVM)的执行能力对其进行动态排序,采用改进的M_V_O蚁群算法对带有偏序关系的一系列短作业进行组合寻优调度,考虑到了云计算软件定义网络中的延时等因素局部更新蚂蚁的信息素浓度,并通过全局正向反馈增强最优解的收敛速度。本文理论上分析了该算法的有效性,且在CloudSIM平台下通过实验验证了该算法的可行性和有效性。
A long job can be split into a series of short ones on distributed computing environment.And taking optimal scheduling strategy to the divided short jobs,the turnaround time and system response time of the long job is shortened.Being directed against the short comings of traditional scheduling policies and the characteristics of network delay in cloud computing,the resource in the virtual level is reallacated with the KVM web hosts.The KVM web hosts was sorted dynamically by the customized operation capability,and modified M_V_0ant colony al-gorithms was taken to do task scheduling combinatorial optimization for the short jobs with partial orders,and local-ly updated the thickness of pheromones considered the delaying factors in the network,and accelerated optimization convergence speed by global positive feedback.The validity of the algorithms was analyzed theoretically,and the performance and feasibility of the algorithms under the platform of Cloud SIM were verified.
作者
王灿伟
WANG Can-wei(Department of Computer Science and Technology, Nanjing University, Nanjing City 210046 , P. R. China;Department of Information and Engineering, Shandong Management University, Jinan 250357 , P. R. China)
出处
《科学技术与工程》
北大核心
2016年第30期115-121,共7页
Science Technology and Engineering
基金
国家自然科学基金青年项目(71301086)
山东省电子政务项目(2150511)
山东省科技厅星火计划项目(2013XH17003)
教育厅科技计划项目(J14LN62)资助
关键词
云计算
虚拟化
任务调度
蚁群算法
cloud computing
virtualization
task scheduling
ant colony algorithms