云环境下能耗感知的虚拟机动态迁移策略研究

发布时间：2018-04-24 08:17

本文选题：动态电压频率调节 + 服务等级协定违例　；参考：《微电子学与计算机》2015年09期

【摘要】：针对云计算服务环境下静态阈值的虚拟机(Virtual Machine,VM)动态迁移算法无法根据云任务负载实时优化系统能耗和负载均衡的问题,(方法)提出利用工作负载历史数据的统计学规律自适应地确定迁移阈值,实现云计算环境下虚拟机(Virtual Machine,VM)动态调度的自适应节能算法.(结果)采用动态电压频率调节(dynamic voltage frequency scaling,DVFS)--实现系统部件静态节能,又通过两种自适应主机超载判定(Adaptive Host Overloading Detection,AHOD)算法实现云数据中心的VM动态迁移,在Cloudsim云仿真平台下对比实现DVFS静态节能和2种AHOD动态节能策略,(结论)结果表明:DVFS和AHOD策略可以显著节能68%以上,虽然AHOD策略中VM动态迁移引起的服务等级协定违例(Service Level Agreement Violation,SLAV)将导致云服务质量(Quality of Service,QoS)降低,并造成数据中心综合性能下降,然而合理建模并找到能耗与QoS的平衡点,即安全阈值为1.2的AHOD算法不但能够实现自适应负载均衡的虚拟机动态迁移,还实现云环境下的高效节能.
[Abstract]:Virtual Machine VM-based dynamic migration algorithm for static threshold in cloud computing service environment can not optimize system energy consumption and load balance according to cloud task load in real time. The law adaptively determines the threshold of migration, An adaptive energy saving algorithm for dynamic scheduling of Virtual Machine (VM) in cloud computing environment is implemented. (results) dynamic voltage frequency scaling is used to realize static energy saving of system components, and the VM dynamic migration of cloud data center is realized by two adaptive Host Overloading detection algorithm. The comparison of DVFS static energy saving and two AHOD dynamic energy-saving strategies (conclusion) based on Cloudsim cloud simulation platform shows that DVFS and AHOD strategies can save more than 68% energy significantly. Although the service level agreement violation caused by VM dynamic migration in the AHOD policy will lead to a decrease in the quality of service quality of service QoS and the deterioration of the comprehensive performance of the data center, the balance between energy consumption and QoS will be found. That is, the AHOD algorithm with a security threshold of 1. 2 can not only realize the dynamic migration of the adaptive load balancing virtual machine, but also realize the high efficiency and energy saving in the cloud environment.
【作者单位】：新疆大学软件学院;
【基金】：国家自然科学基金项目(61363083,61262088) 新疆青年博士科技人才创新项目(2013731004) 新疆高校科研计划项目(XJEDU2012I10) 新疆自治区自然科学基金(2013211A011,2011211A011)
【分类号】：TP302

【参考文献】