开源云管理平台OpenStack中虚拟机部署机制的研究与优化

发布时间：2018-06-02 03:34

本文选题：OpenStack + 互联网　；参考：《北京邮电大学》2013年硕士论文

【摘要】：近年来,云计算一词越来越多的出现在人们的视野当中,也获得了越来越多的关注。云计算描述了一种基于互联网的新的IT服务增加、使用和交付模式,通常涉及通过互联网来提供动态可扩展而且经常是虚拟化的资源。云计算所带来的不仅仅是技术上的变革,更是商业模式上的变革。从长远来看,伴随着商业模式的变革,无论是企业用户还是个人用户,都会大幅降低投入成本,提高生产效率。借助云计算,社会各个领域会更廉价、更快捷的获取IT服务。作为现在最炙手可热的IT技术,云计算正在并将继续改变整个IT产业的格局。可以预见,云计算将像几十年前的个人电脑和互联网一样引领IT行业新一轮的浪潮。然而,无论云计算怎么发展,最终都需要依托数据中心来对外提供服务,数据中心尤其是采用虚拟化技术的数据中心的受关注程度也越来越高。同时,传统数据中心正在消耗大量的能源、空间和成本,而且消耗量日益膨胀。随着信息服务的日益深化,越来越多的企业希望构建更加绿色更加节能的新型数据中心,提高数据中心各项资源的利用率和节能降耗已成为当前数据中心面临的首要问题,相应的虚拟机部署问题也成为IT领域的重要研究方向。近两年来,通过虚拟化手段进行集中化整合开始成为提高数据中心利用率且节能的有力途径。部署了虚拟化技术之后,我们需要利用云计算系统的平台管理技术来实现大量服务器的协同工作,帮助我们便捷的进行业务部署和开通,快速发现和恢复系统故障,通过自动化、智能化的手段实现大规模系统的可靠运营。随着云计算的发展,云计算系统的平台管理技术也越来越成熟,OpenStack就是其中的佼佼者。但是其虚拟机部署机制还不太完善,存在如下问题：第一,采用原虚拟机部署算法的时候,使用状态的物理机的各项资源利用率偏低,造成了资源浪费；第二,原虚拟机部署算法不能为每一台物理机找到最合适的虚拟机组合,从而使用的物理机的数量偏多,增加了不必要的电能损耗,也增加了运营成本。本文针对新兴开源云计算平台OpenStack中虚拟机部署机制的不足,设计实现了一项新的解决方案——基于蚁群算法的OpenStack虚拟机部署策略。该策略利用了蚂蚁寻找食物过程中的分布式协作和正反馈机制来处理虚拟机请求,以减少使用的物理机数量并且降低电能消耗。通过搭建真实实验环境,验证了采用新虚拟机部署策略的OpenStack管理平台的可用性和可靠性,然后通过仿真实验的方式比较了改进算法和原来算法在各个层面的性能。证明了改进算法可以提高处于使用状态的物理机的各项资源的利用率,并能减少同等情况物理机使用数量。
[Abstract]:In recent years, the word cloud computing appears more and more in the field of vision, and also gets more and more attention. Cloud computing describes a new Internet-based IT service growth, usage, and delivery model, often involving the provision of dynamically scalable and often virtualized resources via the Internet. Cloud computing brings not only technological changes, but also business model changes. In the long run, with the change of business model, both enterprise users and individual users will greatly reduce the input cost and improve production efficiency. With cloud computing, all sectors of society will be cheaper, faster access to IT services. As the hottest IT technology, cloud computing is and will continue to change the pattern of the entire IT industry. Cloud computing is expected to lead the IT industry into a new wave, as personal computers and the Internet did decades ago. However, no matter how cloud computing develops, it will ultimately need to rely on data centers to provide services to the outside world. Data centers, especially those using virtualization technology, are attracting more and more attention. At the same time, traditional data centers are consuming a lot of energy, space and costs, and consumption is expanding. With the deepening of information services, more and more enterprises want to build new data centers that are more green and more energy efficient. The corresponding virtual machine deployment problem has also become an important research direction in IT field. In the past two years, centralized integration through virtualization has become a powerful way to improve data center utilization and save energy. After the deployment of virtualization technology, we need to use cloud computing platform management technology to achieve a large number of servers to work together, help us to easily deploy and open business, quickly detect and recover system failures. The reliable operation of large-scale system is realized by means of automation and intelligence. With the development of cloud computing, the platform management technology of cloud computing system is more and more mature. OpenStack is one of the best. However, the mechanism of virtual machine deployment is not perfect, there are some problems as follows: first, when using the original virtual machine deployment algorithm, the utilization ratio of each resource of the state physical machine is low, resulting in a waste of resources; second, The original virtual machine deployment algorithm can not find the most suitable combination of virtual machines for each physical computer, so the number of physical machines used is on the high side, which increases the unnecessary power consumption, and also increases the operating cost. Aiming at the shortage of virtual machine deployment mechanism in OpenStack, a new solution, OpenStack virtual machine deployment strategy based on ant colony algorithm, is designed and implemented in this paper. The strategy takes advantage of the distributed collaboration and positive feedback mechanism in the process of searching for food to process virtual machine requests to reduce the number of physical machines used and the power consumption. The availability and reliability of the OpenStack management platform using the new virtual machine deployment strategy are verified by building a real experimental environment. Then the performance of the improved algorithm and the original algorithm at various levels are compared by simulation experiments. It is proved that the improved algorithm can increase the utilization rate of all kinds of resources and reduce the number of physical computers in the same situation.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP302

【相似文献】