高可用性虚拟化管理框架的研究与实现
发布时间：2018-12-14 02:57
【摘要】：最近几年,越来越多的用户考虑建设虚拟化管理中心。究其原因,主要有三点：硬件成本降低、空间更节省、能耗更环保。这样不仅可以帮助用户把应用从物理服务器迁移到虚拟机(VM)上运行,还可以提供基于硬件层面的能耗优化和负载分担机制,从而提高计算资源的利用效率。 虚拟化的高可用性应用为其提供了更高的平台空间,其中以高可用性服务器集群为典型代表。高可用性服务器集群化可以确保当一台服务器发生故障时,虚拟机会自动地在另一台备用服务器上重启,而无需破坏用户的当前操作。集群由两台或者多台配置相同的服务器(物理机或者虚拟机均可)组成。集群软件确保应用程序数据在所有服务器上的更新一致,并且能够在主服务器发生故障时,重启备用服务器。 作为虚拟化管理框架的高标准要求,高可用性一直是研究的热点。本文在Convirt二次开发软件NEU-Convirture的基础上,基于RHCS (Red Hat Cluster Suite)集群套件中的LVS(基于Linux的虚拟服务器集群技术)提供负载均衡功能,以提供虚拟化管理框架的高可用性。本文还针对LVS提供的几种负载均衡算法进行比较,从而提出了一种新的负载均衡算法来改善系统的性能,使多种资源可以在多台计算机之间进行合理的分配,基本达到均衡一致,避免某些节点过于清闲,而某些节点负载较重的问题。 首先,本文通过对Convirture以及二次开发软件NEW_Convirture的研究分析,搭建高可用性虚拟化管理框架的应用平台,确保对虚拟存储、虚拟化服务器、集群、虚拟机等虚拟对象的操作可行性与可靠性,以完成基本的功能需求。 其次,本文通过分析集群套件RHCS的结构以及核心功能,重点剖析实现其中涉及的调度算法,主要包括加权循环调度算法(WRRS),最小连接调度算法(LCS),加权最小连接调度算法(WLCS)等,然后提出改进的负载均衡调度算法,以提高系统的性能,确保高可用性虚拟化管理框架的可行、可用、可靠。 最后,本文通过搭建的实验环境,完成实现上述基本功能的操作和负载均衡算法性能的测试对比分析。通过对虚拟化管理框架的实验测试可以看出,该算法可以实现系统的高可用性,提供一个稳定良好的运行环境。
[Abstract]:In recent years, more and more users are considering building virtualization management centers. There are three main reasons: lower hardware cost, more space saving, and more environmental protection. This can not only help users migrate their applications from physical servers to virtual machine (VM), but also provide hardware based energy optimization and load-sharing mechanisms to improve the efficiency of computing resources. Virtualization provides higher platform space for high availability applications, which are typically represented by high availability server clusters. High availability server clustering ensures that when one server fails, the virtual machine automatically restarts on another standby server without disrupting the user's current operation. The cluster consists of two or more servers (physical or virtual machines) with the same configuration. The cluster software ensures that the application data is updated consistently on all servers and that the standby server can be restarted if the primary server fails. As a high standard requirement of virtualization management framework, high availability is always a hot research topic. On the basis of Convirt secondary development software NEU-Convirture and LVS (Virtual Server Cluster Technology based on Linux) in RHCS (Red Hat Cluster Suite) cluster suite, this paper provides load balancing function to provide high availability of virtualization management framework. This paper also compares several load balancing algorithms provided by LVS, and puts forward a new load balancing algorithm to improve the performance of the system, so that a variety of resources can be allocated reasonably among multiple computers. The balance is basically consistent to avoid the problem that some nodes are too idle and some nodes are heavily loaded. First of all, through the research and analysis of Convirture and secondary development software NEW_Convirture, this paper builds the application platform of high availability virtualization management framework to ensure the virtual storage, virtualization server, cluster, virtual storage, virtual server, cluster. The operation feasibility and reliability of virtual objects, such as virtual machines, can meet basic functional requirements. Secondly, by analyzing the structure and core functions of cluster suite RHCS, this paper focuses on the implementation of scheduling algorithms involved, including the weighted cyclic scheduling algorithm (WRRS), minimum connection scheduling algorithm (LCS),. The weighted minimum connection scheduling algorithm (WLCS) is proposed, and then an improved load balancing scheduling algorithm is proposed to improve the performance of the system and ensure the feasibility, availability and reliability of the high availability virtualization management framework. Finally, through the experimental environment, the operation of the above basic functions and the performance of load balancing algorithm are compared and analyzed. The experimental test of the virtualization management framework shows that the algorithm can achieve high availability of the system and provide a stable and good running environment.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP302
本文编号：2377779
[Abstract]:In recent years, more and more users are considering building virtualization management centers. There are three main reasons: lower hardware cost, more space saving, and more environmental protection. This can not only help users migrate their applications from physical servers to virtual machine (VM), but also provide hardware based energy optimization and load-sharing mechanisms to improve the efficiency of computing resources. Virtualization provides higher platform space for high availability applications, which are typically represented by high availability server clusters. High availability server clustering ensures that when one server fails, the virtual machine automatically restarts on another standby server without disrupting the user's current operation. The cluster consists of two or more servers (physical or virtual machines) with the same configuration. The cluster software ensures that the application data is updated consistently on all servers and that the standby server can be restarted if the primary server fails. As a high standard requirement of virtualization management framework, high availability is always a hot research topic. On the basis of Convirt secondary development software NEU-Convirture and LVS (Virtual Server Cluster Technology based on Linux) in RHCS (Red Hat Cluster Suite) cluster suite, this paper provides load balancing function to provide high availability of virtualization management framework. This paper also compares several load balancing algorithms provided by LVS, and puts forward a new load balancing algorithm to improve the performance of the system, so that a variety of resources can be allocated reasonably among multiple computers. The balance is basically consistent to avoid the problem that some nodes are too idle and some nodes are heavily loaded. First of all, through the research and analysis of Convirture and secondary development software NEW_Convirture, this paper builds the application platform of high availability virtualization management framework to ensure the virtual storage, virtualization server, cluster, virtual storage, virtual server, cluster. The operation feasibility and reliability of virtual objects, such as virtual machines, can meet basic functional requirements. Secondly, by analyzing the structure and core functions of cluster suite RHCS, this paper focuses on the implementation of scheduling algorithms involved, including the weighted cyclic scheduling algorithm (WRRS), minimum connection scheduling algorithm (LCS),. The weighted minimum connection scheduling algorithm (WLCS) is proposed, and then an improved load balancing scheduling algorithm is proposed to improve the performance of the system and ensure the feasibility, availability and reliability of the high availability virtualization management framework. Finally, through the experimental environment, the operation of the above basic functions and the performance of load balancing algorithm are compared and analyzed. The experimental test of the virtualization management framework shows that the algorithm can achieve high availability of the system and provide a stable and good running environment.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP302
【参考文献】
相关期刊论文 前6条
1 何锋;Linux下Java程序的编译与调试[J];电脑编程技巧与维护;2002年09期
2 龚梅;王鹏;吴跃;;一种集群系统的透明动态反馈负载均衡算法[J];计算机应用;2007年11期
3 陈宣;罗军;谭郁松;李姗姗;;集群系统中自适应负载反馈平衡策略的研究[J];计算机应用与软件;2006年08期
4 陈康;郑纬民;;云计算:系统实例与研究现状[J];软件学报;2009年05期
5 刘爱军;耿国华;;基于x86的虚拟机技术现状、应用及展望[J];计算机技术与发展;2007年11期
6 李欣;孙剑虹;;服务器虚拟化技术在系统集成中的应用[J];科技资讯;2007年33期
，本文编号：2377779
本文链接：https://www.wllwen.com/kejilunwen/jisuanjikexuelunwen/2377779.html
