毕业论文-电信云下硬件资源调度的算法研究.docx

浙江大学硕士学位论文STYLEREF 标题,章标题(无序号) \* MERGEFORMATAbstract PAGE i 电信云下硬件资源调度的算法研究 摘要 当前云计算技术飞速发展，云服务大量地向公众、向企业用户开放，而在需要大量数据处理的电信领域，云计算的发展必然不可阻挡。随着云规模的不断扩大，如何优化提升云计算系统的性能成为了一个难题。本文从云系统内资源调度的角度来实现其性能的调优。针对电信云计算数据中心的特征，本文从数学角度构建了它的模型，并且实现了一个初始放置算法用以将虚拟机部署到电信云计算数据中心之中。 电信云计算数据中心的建模主要从数据中心的物理资源、网络资源、虚拟资源三个方面出发，在准确描绘出这些特征的同时，联系电信云资源调度系统的优化目标给出相应的约束条件和优化目标函数。在模型的基础上，本文设计并实现了电信云资源调度系统中的初始放置算法，包含了物理主机选择、虚拟机分组、虚拟机放置三个阶段。该算法在保证虚拟机通信时延小于规定值的情况下，提升系统性能降低电信云数据中心网络中的整体流量。 当然，本文存在不少不足之处，例如针对资源调度中的负载均衡、热点消除等需求并没有给出解决方案，算法对性能的优化程度及执行效率仍有待提升，这些都需要在未来的工作中使之实现。 关键词：　云计算，资源调度，建模，初始放置 浙江大学硕士学位论文 Abstract 浙江大学硕士学位论文Abstract PAGE ii PAGE ii Abstract In current days, cloud computing technology develops in a high speed, the cloud service is open to the public and the enterprise users in large quantities. With the enlargement of the scale of cloud, the optimization of the performance of cloud computing system has become a difficult problem. According to the characteristics of the telecom cloud computing data center, its mathematical model had been built in this paper, and an algorithm designed for initial placement in data center to deploy virtual machines. The modeling of data center starts from three aspects: physical resources, network resources and virtual resources. While describing these kinds of characteristics accurately, the constraints and optimization functions are also given to achieve the optimization goals of telecom cloud resource scheduling system. This paper has completed the designation and implement an initial placement algorithm, it includes three phases: the selection of physical machines, the partition of virtual machines and the placement of virtual machines. The algorithm ensures the delay of communication between any two virtual machines, further, it also improves the performance of cloud computing system by means of reducing the whole communication volumes in data center. Of course, it still contains some lacks in th