基于云平台的虑拟机在线迁移技术研究
发布时间:2018-04-24 10:18
本文选题:云计算 + OpenStack ; 参考:《南京邮电大学》2017年硕士论文
【摘要】:能源消耗量是云计算中心重点考虑的因素之一,然而在现有云计算环境中,服务的动态扩展导致云计算中心中某些节点负载过小,在造成资源闲置的同时带来能源浪费问题;此外,现有的云平台中虚拟机迁移技术无法同时保证“不间断”服务以及高鲁棒性。因此有必要研究云计算环境中针对节能问题的虚拟机在线迁移技术。本文主要研究面向节能的虚拟机在线迁移技术。针对现有节点负载率性能参数单一问题,提出一种可针对不同应用的更灵活的节点负载率计算方法;针对节能目标,综合考虑虚拟机的迁移数量因素,提出一种资源调度规则,根据规则选择需要迁移的虚拟机,为每个待迁移的虚拟机匹配最佳节点;针对现有迁移技术无法同时保证“不间断”服务和高鲁棒性问题,提出一种基于内存页混合拷贝方式的虚拟机在线迁移技术,该技术在保证“不间断”服务的同时,通过优先从本地获取内存页的方式,降低内存页丢失的风险,保证鲁棒性,并减少内存页传输的时间,降低虚拟机迁移总时间;在所有迁移工作完成之后,关闭零负载节点,达到节能的目的。基于OpenStack开源云平台实现了面向节能的虚拟机在线迁移原型系统,在原型系统上进行能源节约效率和虚拟机在线迁移性能的测试。测试结果表明:基于OpenStack的面向节能的虚拟机在线迁移技术,能够降低虚拟机在线迁移过程中对服务质量的影响,并实现节能的优化目标。
[Abstract]:Energy consumption is one of the key factors considered by cloud computing center. However, in the current cloud computing environment, the dynamic expansion of services leads to a small load of some nodes in the cloud computing center, which results in energy waste while the resources are idle. In addition, the existing virtual machine migration technology in the cloud platform can not guarantee "continuous" service and high robustness at the same time. Therefore, it is necessary to study virtual machine online migration technology for energy saving in cloud computing environment. This paper focuses on energy-saving virtual machine online migration technology. Aiming at the problem of single node load rate performance parameter, a more flexible node load rate calculation method for different applications is proposed, and a resource scheduling rule considering the migration quantity factor of virtual machine is proposed for the purpose of energy saving. According to the rules, we select the virtual machines that need to be migrated, and match the best nodes for each virtual machine to be migrated. In view of the problem that the existing migration technology can not guarantee "uninterrupted" service and high robustness at the same time, This paper presents a virtual machine online migration technology based on the mixed copy of memory pages. This technology ensures "continuous" service while reducing the risk of loss of memory pages and ensuring robustness by prioritizing local access to memory pages. It also reduces the time of page transfer, reduces the total migration time of virtual machine, closes the zero load node after all migration work is finished, and achieves the purpose of saving energy. Based on OpenStack open source cloud platform, a prototype system of virtual machine online migration for energy saving is implemented. The efficiency of energy saving and the performance of virtual machine online migration are tested on the prototype system. The test results show that the energy saving virtual machine online migration technology based on OpenStack can reduce the impact of virtual machine online migration on the quality of service and achieve the optimization goal of energy saving.
【学位授予单位】:南京邮电大学
【学位级别】:硕士
【学位授予年份】:2017
【分类号】:TP302;TP393.09
【参考文献】
相关期刊论文 前2条
1 张慧珍;李长春;周培琴;赵利红;;云计算数据中心节能技术研究[J];信息系统工程;2013年10期
2 谭一鸣;曾国荪;王伟;;随机任务在云计算平台中能耗的优化管理方法[J];软件学报;2012年02期
相关硕士学位论文 前2条
1 陈恂;OpenStack平台的虚拟机调度关键技术研究[D];杭州电子科技大学;2014年
2 王旭;基于能耗约束的虚拟机调度策略研究[D];云南大学;2011年
,本文编号:1796233
本文链接:https://www.wllwen.com/kejilunwen/jisuanjikexuelunwen/1796233.html