云环境下面向节能的虚拟机调度研究

发布时间：2018-01-12 19:21

本文关键词：云环境下面向节能的虚拟机调度研究　出处：《华东师范大学》2017年硕士论文　论文类型：学位论文

【摘要】：近年来随着越来越多的服务和应用迁至云端运行,云服务提供商不得不扩大云数据中心的规模。大规模的数据中心不仅消耗巨大的电能,同时释放的大量二氧化碳也会对全球气候造成一定的影响。因此,近年来数据中心节能技术受到了各界的广泛关注和研究。为保证云服务提供商自身的利益和云用户的满意度,用户在使用云服务时需要和云服务提供商签订服务等级协议(SLA,Service-level Agreement)。本文主要研究如何在保证尽量不违反SLA协议的前提下通过对虚拟机的合理调度达到数据中心节能的目的。为研究如何降低数据中心能耗,本文首先分析了虚拟机调度的系统模型,分析了虚拟机调度中待迁移虚拟机的选取及放置问题的数学模型,分析了云数据中心中单台服务器的能耗模型及虚拟机迁移过程中产生的能耗开销和由于违反SLA所带来的开销,得出通过设置合理的CPU使用率阈值,可以有效降低总开销。在本文的第四部分,我们分析了使用静态阈值和动态自适应阈值检测过载服务器的优缺点及适用场景;分析了贪心算法选择待迁移虚拟机集合的缺点并提出了基于动态规划的虚拟机选择方法;在待迁移虚拟机的放置问题上,我们分析了基于降序最佳适应算法来解决待迁移虚拟机放置问题的算法,并提出运用人工鱼群算法来解决待迁移虚拟机放置的方法。通过实验证明本文所提出的虚拟机选择算法和虚拟机放置算法具有较好的节能效果,并且能够保证SLA违反率在可接受范围内。
[Abstract]:In recent years, with more and more services and applications moving to the cloud, cloud service providers have to expand the scale of cloud data centers. The amount of carbon dioxide released at the same time will also have an impact on the global climate. In recent years, data center energy-saving technology has been widely concerned and studied. In order to ensure cloud service providers' own interests and cloud users' satisfaction. Users need to sign a service level agreement (SLA) with the cloud service provider when using the cloud service. Service-level agreements). This paper mainly studies how to achieve the goal of data center energy saving by reasonable scheduling of virtual machine without violating SLA protocol as far as possible, and how to reduce the energy consumption of data center in order to study how to reduce the energy consumption of data center. This paper first analyzes the system model of virtual machine scheduling, and analyzes the mathematical model of the selection and placement of the virtual machine to be migrated in the virtual machine scheduling. The energy consumption model of a single server in the cloud data center, the energy consumption incurred during the migration of virtual machine and the overhead caused by violating SLA are analyzed, and the reasonable threshold of CPU utilization is obtained. In the 4th part of this paper, we analyze the advantages and disadvantages of using static threshold and dynamic adaptive threshold to detect overload server. The shortcomings of greedy algorithm in selecting the set of virtual machines to be migrated are analyzed and a dynamic programming based virtual machine selection method is proposed. On the placement of virtual machines to be migrated, we analyze the algorithm based on descending optimal adaptation algorithm to solve the problem of virtual machines to be migrated. An artificial fish swarm algorithm is proposed to solve the problem of virtual machine placement. Experiments show that the virtual machine selection algorithm and virtual machine placement algorithm proposed in this paper have a good energy-saving effect. And can ensure that the SLA violation rate is within acceptable range.
【学位授予单位】：华东师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP302

【参考文献】