云环境下虚拟机部署技术的研究及实现

发布时间：2018-01-07 12:45

  本文关键词：云环境下虚拟机部署技术的研究及实现　出处：《东北大学》2013年硕士论文　论文类型：学位论文

  更多相关文章： Xen 虚拟机 部署技术 云计算 遗传算法

【摘要】：近年来,云计算技术受到了广泛的关注,是当前学术界、产业界的研究热点。而虚拟化技术由于具有增强云计算系统弹性和扩展性、提高资源使用效率、减少硬件投资及维护成本等诸多优势而被广泛的使用,已成为当前云计算领域的关键技术。虚拟机部署技术是虚拟化领域的研究热点之一,主要聚焦于采用合适的策略和算法,将虚拟机快速部署到物理机上,并使虚拟机在集群上高效运行。现有的部署算法主要针对CPU等单一因素而提出,而且通常没有考虑虚拟机负载的类型,因而存在着一定的局限性；另外,现有的基于镜像方式的虚拟机快速部署方式还存在着用户满意度低和管理复杂两方面的问题。针对上述问题,本文在Xen虚拟化平台的基础上,具体做了如下工作：(1)针对现有虚拟机部署策略和算法存在的考虑因素单一和目标不明确的问题,提出了两种虚拟机部署算法,即基于矩阵加权的虚拟机部署算法(MW-EC)和基于遗传算法的虚拟机部署算法(GA-LB),前者在考虑集群系统能耗的同时,重点考虑服务质量,能够有效提高服务质量；而后者则是在改进原有遗传算法的基础上设计的,具有很强的负载均衡能力。(2)结合上述两种虚拟机部署算法,设计并实现了一个云环境下的虚拟机部署管理框架VMPF,该框架由虚拟机部署管理模块、部署代理模块、资源监控模块和模板管理模块等组成。部署管理模块能够根据用户请求、按照虚拟机部署算法生成部署方案；部署代理模块则根据部署方案来创建虚拟机,完成部署；资源监控模块主要收集物理机的负载信息,为部署管理模块服务；模板管理模块则提供虚拟机镜像管理功能。该框架能够提供用户便捷而快速地申请部署虚拟机,并实现了用户申请虚拟机自动化,具有很高的部署效率。(3)设计并实现了所提出的虚拟机部署框架和算法,对提出的基于矩阵加权的部署算法和基于遗传算法的虚拟机部署算法进行了性能测试,并和现有FF算法、FFD算法和择优算法进行了比较。实验结果表明基于矩阵加权的虚拟机部署算法能够达到良好的节能目标,而基于遗传算法的虚拟机部署算法具有很好的负载均衡能力。
[Abstract]:In recent years, cloud computing technology has received extensive attention, is the current research focus of academia, industry and virtualization technology has enhanced cloud computing system flexibility and scalability, improve the efficiency of resource use, reduce the hardware investment and maintenance cost advantages and is widely used, has become a key technology in the cloud in the calculation domain. The virtual machine deployment technology is one of the hot research field of virtualization, mainly focusing on the strategy and algorithm of virtual machine rapid deployment to the physical plane, and the efficient operation of the virtual machine in the cluster. The existing deployment algorithms for CPU single factor is put forward, and usually no considering the type of virtual machine load, so there are some limitations; in addition, the existing virtual machine image based on the way of rapid deployment mode still has a low customer satisfaction and complex management Two aspects of the problem. To solve the above problems, based on the Xen virtualization platform, the specific contents are as follows: (1) according to the existing virtual machine deployment strategy and algorithm are simple consideration and vague goals, put forward two kinds of virtual machine deployment algorithm, namely virtual machine based on weighted matrix the deployment algorithm (MW-EC) and virtual machine deployment algorithm based on genetic algorithm (GA-LB), the former in considering the cluster energy consumption at the same time, considering the quality of service, can effectively improve the quality of service; the latter is designed in the improved genetic algorithm based on load balancing, has a very strong ability. (2 two) combined with the virtual machine deployment algorithm, design and implementation of virtual machines in a cloud environment management under the framework of VMPF deployment, the framework of the virtual machine deployment management module, deployment agent module, resource monitoring module and template tube Management module. Deployment management module can according to user requests, in accordance with the virtual machine deployment algorithm to generate deployment scheme; deployment agent module is based on the deployment plan to create a virtual machine, the completion of the deployment of information resources; load monitoring module mainly collects the physical machine, for service deployment management module; template management module provides a virtual machine image management function the framework can provide users convenient and fast for the deployment of virtual machines, and realizes the user application virtual machine automation, has high deployment efficiency. (3) the virtual machine deployment framework and algorithm design and implementation of the proposed, the proposed deployment algorithm based on weighted matrix and the virtual machine deployment algorithm genetic algorithm based on the performance test, and the existing FF algorithm, FFD algorithm and compare the preferred algorithm. The experimental results show that the virtual machine based on weighted matrix The deployment algorithm can achieve good energy saving targets, and the virtual machine deployment algorithm based on genetic algorithm has a good load balancing ability.

【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP302
，

本文编号：1392624