云计算环境下虚拟机资源分配及部署策略研究

发布时间：2018-04-02 17:33

本文选题：云计算　切入点：虚拟机　出处：《云南大学》2012年硕士论文

【摘要】：云计算的出现改变了传统的计算资源交付模式,是当前研究的热点。由于云计算的按需提供资源、按使用量付费以及动态伸缩性等特点,对于云计算的研究还存在着诸多挑战,特别是虚拟机资源的分配、部署、迁移以及物理服务器的整合等。虚拟机资源的分配与部署是云计算研究中的两个核心问题。传统的对于虚拟机资源分配与部署的研究都是基于单方面的考虑,忽略了一些用户和云服务提供商所关心的参数。因此,研究虚拟机资源在云计算市场中如何进行分配以及虚拟机在数据中心内如何部署具有重要的学术意义和应用价值。针对虚拟机资源分配问题,本文以云计算市场中的云计算服务提供商为研究对象,引入了经济学中的演化博弈理论,对云计算中虚拟机资源的分配进行建模分析。在整个云计算市场中,云计算服务提供商发布的虚拟机资源的价格以及服务质量作为博弈策略。对云服务提供商发布的价格和服务质量进行演化博弈,在博弈的过程中,云计算服务提供商为了提高自身的效用动态的调整自己的博弈策略。最后从理论上和实验上证明了整个虚拟机资源分配过程能达到演化稳定。针对虚拟机资源部署问题,本文以云计算服务提供商的数据中心为研究对象,以粒子群优化算法为理论指导,对数据中心内虚拟机资源的部署进行建模分析。对每个物理机和虚拟机的CPU、内存、存储进行量化,根据用户的需求和当前物理服务器的负载状态,采用多目标粒子群优化算法根据物理服务器资源利用率、虚拟机迁移次数这两个目标进行虚拟机的优化部署。通过仿真实验结果表明,该算法在最少的虚拟机迁移次数下能有效的提高资源利用率。
[Abstract]:The emergence of cloud computing has changed the traditional mode of computing resource delivery, which is the focus of current research.Due to the characteristics of on-demand, pay-per-use and dynamic scalability of cloud computing, there are still many challenges in cloud computing research, especially the allocation, deployment, migration and integration of physical servers.The allocation and deployment of virtual machine resources are two core issues in cloud computing research.Traditional researches on virtual machine resource allocation and deployment are based on unilateral considerations, ignoring some parameters concerned by users and cloud service providers.Therefore, it has important academic significance and application value to study how to allocate virtual machine resources in cloud computing market and how to deploy virtual machine in data center.Aiming at the allocation of virtual machine resources, this paper takes cloud computing service providers in the cloud computing market as the research object, and introduces the evolutionary game theory in economics to model and analyze the allocation of virtual machine resources in cloud computing.In the whole cloud computing market, the price of virtual machine resources released by cloud computing service providers and the quality of service as a game strategy.In the process of evolution game between the price and service quality released by cloud service provider cloud computing service provider dynamically adjusts its own game strategy in order to improve its utility.Finally, it is proved theoretically and experimentally that the whole virtual machine resource allocation process can achieve evolutionary stability.Aiming at the problem of virtual machine resource deployment, this paper takes the data center of cloud computing service provider as the research object and the particle swarm optimization algorithm as the theoretical guidance to model and analyze the deployment of virtual machine resources in the data center.The CPU, memory and storage of each physical machine and virtual machine are quantified. According to the demand of the user and the load state of the current physical server, the multi-objective particle swarm optimization algorithm is adopted according to the resource utilization of the physical server.The two targets of virtual machine migration are optimized deployment of virtual machine.The simulation results show that the algorithm can effectively improve the resource utilization under the minimum number of virtual machine migration.
【学位授予单位】：云南大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP302;O225

【参考文献】