随机负载下基于SLA的虚拟机调度与迁移策略研究

发布时间：2018-10-20 07:53

【摘要】：随着云计算用户数量的不断增长,用户市场也随之逐渐细分。云服务提供商针对低端用户群体推出的廉价云服务通常会采用超售的方式售卖云计算资源。虽然通过资源超售能够大幅降低成本,但是当数据中心虚拟机节点的负载产生波动时,会有很大几率导致云服务质量的下降,增加违反与用户签订的SLA协议的风险。为了在对资源进行超售的同时保证云服务的质量,云服务提供商会使用虚拟机迁移技术对数据中心内的资源进行动态调整。随着云计算规模的不断扩大以及云计算业务的全球化,数据中心内的虚拟机负载波动变得难以寻找规律。如何在虚拟机随机负载波动情况下保证云服务的质量便成为了需要攻克的难关。针对以上问题,本文从以下三个方面开展研究工作。(1)在对云计算、虚拟化的相关概念及研究现状进行梳理的基础上,阐述了虚拟机的生命周期;围绕云计算服务质量展开讨论,对影响服务质量的因素及云计算服务质量指标进行归纳总结。(2)研究了随机负载波动环境下,双阈值虚拟机迁移策略中不同虚拟机选择策略与虚拟机放置策略的组合对云计算服务质量的影响。实验结果表明,双阈值虚拟机迁移策略在虚拟机负载波动平稳的环境中能够起到降低虚拟机迁移次数、提高云服务质量的效果,但在随机负载波动环境中表现不佳。使用降序首次适应策略作为虚拟机放置策略,在资源超售的云计算环境中表现优异。使用最大增长策略作为虚拟机选择策略可以使云平台中虚拟机迁移次数降低,而使用最大需求优先策略则可以提高云服务的质量。(3)在双阈值虚拟机迁移策略的基础上进行改进。针对随机负载波动环境提出了一种基于上阈值、分配阈值、下阈值的三阈值虚拟机迁移策略,并对三阈值虚拟机迁移策略中的阈值设定进行研究。实验结果表明,三阈值虚拟机迁移策略在随机负载波动环境中,相比双阈值虚拟机迁移策略能够有效减少虚拟机迁移次数并提高云计算服务质量。
[Abstract]:With the increasing number of cloud computing users, the user market is gradually subdivided. Low-cost cloud services offered by cloud service providers for low-end users typically sell cloud computing resources on an oversold basis. Although overselling of resources can greatly reduce the cost, when the load of the virtual machine node in the data center fluctuates, there is a great chance that the quality of service of the cloud will decline, increasing the risk of violating the SLA protocol signed with the user. In order to oversell resources and ensure the quality of cloud services, cloud service providers use virtual machine migration technology to dynamically adjust the resources in the data center. With the expansion of cloud computing scale and the globalization of cloud computing business, the fluctuation of virtual machine load in data center becomes difficult to find a regular pattern. How to guarantee the quality of cloud service under the random load fluctuation of virtual machine becomes a difficult problem. Aiming at the above problems, this paper carries out research work from the following three aspects. (1) on the basis of combing the related concepts and research status of cloud computing and virtualization, this paper expounds the life cycle of virtual machine, and discusses the quality of service in cloud computing. The factors that affect the quality of service and the quality of service index of cloud computing are summarized. (2) the random load fluctuating environment is studied. The effect of the combination of different virtual machine selection strategy and virtual machine placement strategy on the quality of service of cloud computing in dual-threshold virtual machine migration strategy. The experimental results show that the dual-threshold virtual machine migration strategy can reduce the number of virtual machine migration and improve the cloud quality of service in the environment where the virtual machine load fluctuates smoothly, but it does not perform well in the random load fluctuating environment. The first adaptation strategy of descending order is used as a virtual machine placement strategy, which performs well in the cloud computing environment where resources are oversold. Using the maximum growth strategy as the virtual machine selection strategy can reduce the number of virtual machine migration in the cloud platform, while using the maximum demand first strategy can improve the quality of cloud service. (3) based on the dual-threshold virtual machine migration strategy, we can improve the virtual machine migration. A migration strategy of virtual machine with three thresholds based on the upper threshold, the allocation threshold and the lower threshold is proposed for the random load fluctuation environment, and the threshold setting in the migration strategy of the three threshold virtual machines is studied. The experimental results show that the migration strategy of three-threshold virtual machine can effectively reduce the number of virtual machine migrations and improve the quality of service of cloud computing compared with the dual-threshold virtual machine migration strategy in the random load fluctuating environment.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP302

【相似文献】