云计算中基于XEN的虚拟机动态迁移的研究

发布时间：2018-03-04 11:50

本文选题：云计算　切入点：虚拟化　出处：《武汉理工大学》2013年硕士论文　论文类型：学位论文

【摘要】：云计算描述了一种基于互联网的新的IT服务增加、使用和交付模式,通常书籍通过互联网来提供动态易扩展而且经常是虚拟化的资源。云计算的服务模式分为三种：软件即服务,平台即服务,基础架构即服务。而云计算的部署模型也分为三种：公共云,私有云和混合云。从2006年亚马逊推出弹性计算云服务以来,云服务蓬勃发展,各大IT企业、研究机构都表现出强劲的意愿。本论文将虚拟机动态迁移关键技术的优化作为研究的主要目标。本论文通过对虚拟化中虚拟机动态迁移技术和云计算的全面分析与研究,以现有的Xen虚拟机动态迁移实现机制的分析为基础,重点剖析了虚拟机Xen迁移实现的典型范例,并指出它的优良特征与不足。针对其不足提出Magician(魔术师)动态迁移优化框架。该框架是在典型xen虚拟机运行迁移模块中添加了两个子模块,一个是分层拷贝模块,另一个是内存压缩模块。描述了这两个模块的设计思路,提供了实现该框架的流程图。通过分析,Magician优化了动态迁移实现的时间和空间复杂性。为了验证Magician优化框架的合理性,搭建了实验平台,测试了所设计的优化框架。第一阶段,在零负载情况下,将分层拷贝算法编译到Xen的源代码中,分别在高脏页率和低脏页率环境中,测试实施动态迁移的时间性能。第二阶段,在不同负载情况下,在Xen的优化源代码中增加了内存压缩算法,测试实施动态迁移时的数据量。实验结果表明,本论文提出的分层拷贝算法和内存压缩算法,在虚拟机动态迁移的时间性能上有了良好的改进,能有效缩短迁移宕机时间,实现虚拟机迁移性能优化的目的。本文得到了国家自然科学基金项目(批准号：60970064)的资助。
[Abstract]:Cloud computing describes a new Internet based IT service increased, the use and delivery mode, usually books through the Internet to provide dynamically scalable and often virtualized resources. Cloud computing service model is divided into three types: software as a service, SaaS, IAAs and cloud deployment model. The calculation is divided into three types: public cloud, private cloud and hybrid cloud. Since the launch of the Amazon Elastic Compute Cloud Service from 2006, the cloud service is booming, the major IT companies, research institutions have shown a strong desire.
This paper will optimize the dynamic migration of virtual machines in key technology as the main objective of the study. This paper through a comprehensive analysis and Research on the calculation of the dynamic migration of virtual machines and cloud virtualization technology in the migration mechanism based on dynamic analysis of the existing Xen virtual machine, analyzes the typical example of virtual machine migration realization Xen, and points out its merits and shortcomings. According to the shortcomings of Magician (Magic) dynamic migration optimization framework. The framework is in a typical Xen virtual machine running migration module was added in the two sub modules, a layered copy module and the other is a memory compression module. Describes the design methods of the two modules the flow chart provides an implementation of the framework. Through the analysis of Magician, optimize the time and space complexity of dynamic migration realization.
In order to verify the rationality of the Magician optimization framework, an experimental platform is built, the design optimization framework was tested. In the first stage, the zero load condition, the layered copy algorithm is compiled to Xen source code, respectively, in the high dirty pages dirty page rate and low rate environment, time performance test of dynamic migration. In the second stage, under different load conditions, in the optimization of Xen source code increases the memory test data compression algorithm, implementation of dynamic migration quantity. Experimental results show that the proposed layered copy algorithm and compression algorithm in the virtual memory, have a good time performance improved machine dynamic migration, can shorten the migration downtime, transfer of the performance optimization of the virtual machine.
This paper has been supported by the National Natural Science Fund Project (approval number: 60970064).

【学位授予单位】：武汉理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP302

【参考文献】