摘要
在虚拟机环境下如何有效的管理各类硬件资源,满足用户按需构建和动态配置的要求,当前还是一个新的研究热点。目前主要的资源配置方式主要有两种:虚拟机资源池和资源静态配置。但是,目前针对虚拟机运行状态中的资源动态配置并没有相应的研究。因此,多虚拟机资源动态配置机制的研究具有重要意义。
多虚拟机资源动态配置系统提供了一种虚拟机的集群使用平台,采用资源配置自适应的机制,并结合资源配置的AIMD原则实现了对虚拟机资源的动态配置。多虚拟机资源动态配置系统主要由三部分组成,控制服务层主要的功能是将接收到的虚拟机或者是服务器资源配置的指令转化成为底层物理资源的系统调用,并执行相应的资源调度和配置操作;策略服务层是整个多虚拟机资源动态配置系统的核心部分,主要是依据所收集的平台信息,由用户的需求,启动相关的资源配置策略,生成资源配置的指令,并将此指令向控制服务层传送;用户服务层是面向用户的表示层,通过GUI或者WEB等方式,将整个系统的运行状态,虚拟机和服务器的使用状态等信息,用图形化的方式直观的表现给终端用户。该服务层同样能接受用户手动的资源配置请求,并将该请求向下层组件传送。
通过功能测试和性能测试表明VNIX多虚拟机资源动态配置系统有着良好的功能实现和稳定、高效率的性能表现,同时,在Live-Migrate的应用场景测试中,系统宕机时间缩短了23%左右,特别是在系统高负荷情况下的迁移宕机时间甚至缩短了50%以上。在I/O性能方面,通过虚拟机资源的动态配置优化后,虚拟机I/O的传输速率从22.3MB/S上升至平均31.9MB/S,表明VNIX虚拟机资源动态配置系统对虚拟机的整体运行有着良好的性能提升。
This paper base on the multi-computing system virtualization study, and the current use of cluster laboratory use of the environment, with the development and application technology of virtual machine .Proposed a virtual machine environment of multi-cluster virtual machine management concepts, and completed a prototype system - VNIX. As a virtual technology in Cluster environment , should support high-efficiency, scalable, personalized computing systems dynamic mechanism Construction; establish an user-oriented, application-oriented approach Virtual Computing System; complete the resource adjustable Virtual Machine Manager; according to users or applications of real-time demand, timely and rational allocation and scheduling of server resources. According to VNIX system design and application environment, we proposed the principle of feedback adaptive multi-granularity of information resource management and allocation mechanism. The virtual machine dynamic memory allocation based on the feedback Guest OS dynamic memory allocation strategy based on the tax rate and the Host OS memory optimization strategy, Virtual Machine for the dynamic allocation of CPU virtual machine used for the operational status of coarse-grained adjustment and for each specific VCPU load and practical application of granular optimization strategy.
The function test and the performance test show the success design of the VNIX system. Meanwhile, in Live-Migrate test, system downtime reduced by 23%, especially in high-load, the downtime even short more than 50%. In the I/O performance, through the dynamic allocation of resources, VM I/O transfer speed rises from 22.3 KB / S to average of 31.9 KB/S, that VNIX dynamic resource allocation have a good performance improvements.
引文
[1] James E.Smith,Ravi Nair,“Virtual Machines:Versatile Platforms for Systems and Processes”Mogran Kauffman.2007.113-152
[2] P.R.Goldberg,“Survey of Virtual Machine Research,”Computer,1974.34-35.
[3] James E.Smith,Ravi Nair,“The Architecture of Virtual Machine,”IEEE Computer Society,2005.32-38.
[4] Mendel Rosenblum,Tal Garfinkel,“Virtual Machine Monitors:Current Technology and Future Trends,”IEEE Computer Society,2005.39-47.
[5] X.Jiang and D.Xu,“SODA:A Service-on-Demand Architecture for Application Service Hosting Utility Platforms,”Proc.12th IEEE Int’l Symp.High-Performance Distributed Computing(HPDC-12),IEEE CS Press,2003.174-183.
[6]董耀祖,周正伟,“基于X86架构的系统虚拟机技术与应用”,计算机工程, 2006.32(13):71-73
[7]杨晓伟,廖光灯,胡越明,“基于Xen的X86虚拟机性能调优”,计算机工程,第32卷第24期,2006.32(24):251-253.
[8]毛德操,胡希明.LINUX内核源代码情景分析.浙江:浙江大学出版社,2001.123-208.
[9]李欣,徐飞雪,Xen硬件虚拟机对称多处理器的支持,intel技术开源中心,2007
[10] P.Barham et al.,“Xen and the Art of Virtualization,”Proc.19 The ACM Symp.Operating Systems Principles,ACM Press,2003.164-177
[11] Xen.http://www.xensource.com.
[12] VMware.http://www.vmware.com.
[13] Intel VT http://www.intel.com.
[14] Keir A. Fraser, Steven M. Hand, Ian M. Leslie et al. The XenoServer Computing Infrastructure. Technical Report UCAM-CL-TR-552, University of Cambridge, Computer Laboratory, January 2003. 31~41
[15] Michael Nelson, Beng-Hong Lim, and Greg Hutchins. Fast Transparent Migration forVirtual Machines. In Proceedings of USENIX’05(USENIX 2005), 2005.
[16] VMware virtual center white book.pdf
[17] VMware virtual center user manual.pdf
[18] Carl A.Waldspurger,“Memory Resource Management in VMware ESX Server,”The Symposium on Operating Systems Design and Implementation,2002.181-194
[19] Intel Corp.,“Intel Virtualization Technology Specification for the IA-32 Architecture;”www.intel.com/technology/vt/.
[20] Intel Corp.,“Intel Virtualization Technology Specification for the Intel Itanium Architecture;”www.intel.com/technology/vt/.
[21] Rich Uhlig,Gil Neiger et al.,“Intel Virtualization Technology,”IEEE Computer Society 2005.48-56.
[22] Renato Figueiredo,Peter A.Dinda,Jose Fortes,“Resource Virtualization Renaissance,”IEEE Computer Society 2005.28-31.
[23] Andrew Whitaker,Richard S.Cox et al.,“Rethinking the Design of Virtual Machine Monitors,”IEEE Computer Society 2005.57-62.
[24] P.Ruth,X.Jiang,D.Xu,“Virtual Distributed Environments in a Shared Infrastructure,”IEEE Computer Society 2005.63-69.
[25]杨晓伟,廖光灯,胡越明,“基于Xen的X86虚拟机性能调优”,计算机工程,第32卷第24期,2006.32(24):251-253.
[26]斯大林(Stallings,W)著,魏迎梅等译。操作系统-内核与设计原理。电子工业出版社,2001.6.
[27] Ludmila Cherkasova,Rob Gardner,“Measuring CPU Overhead for I/O Processing in the Xen Virtual Machine Monitor,”USENIX 2005 Annual Technical Conference.
[28] Aravind Menon,Jose Renato Santos et al.,“Diagnosing Performance Overheads in the Xen Virtual Machine Environment,”VEE’05,Chicago,Illinois,USA,2005.
[29] R. Gibbens and F. Kelly. Resource pricing and the evolution of congestion control. Automatica, 35:1969–1985, 1999.
[30] S. Lim, C. Lee, C. Ahn, C. Lee, and K. Park. An adaptive admission control mechanism for a cluster-based web server system. In Proceedings of IPDPS’02, Fort Lauderdale, Florida, April 2002.
[31]深入学习Xen新起点全面介绍Xen虚拟机1-4.http://www.xuniji.com/xen/list.asp,2007.
[32] Alessandro Rubini著,魏永明译。Linux设备驱动程序(第二版)。电力出版社,2004.100-152
[33] K. Ogata. Modern control engineering,. Prentice Hall, 2002.52-67
[34] A. Robertsson, B. Wittenmark, and M. Kihl. Analysis and design of admission control systems in web-server systems. In Proceedings of American Control Conference (ACC), June 2003.
[35] M. Welsh and D. Culler. Adaptive overload control for busy internet servers. In Proceedings of the 4th USENIX Conference on Internet Technologies and Systems,March 2003.
[36] Y. Diao, N. Gandhi, S. Parekh, J. Hellerstein, and D. Tilbury. Using mimo feedback control to enforce policies for interrelated metrics with application to the apache webserver. In Proceedings of the Network Operations and Management Symposium 2002, Florence, Italy, April 2002.
[37] T. Abdelzaher and C. Lu. Modeling and performance control of internet servers. In Proceedings of the 39th IEEE Conference on Decision and Control (ICDC), Sydney, Australia, December 2000.
[38] M. Andersson, M. Kihl, and A. Robertsson. Modelling and design of admission control mechanisms for web servers using non-linear control theory. In Proceedings of ITCom, 2003.
[39]汤凯.netperf与网络性能测量. http://www-128.ibm.com/developerworks/cn/linux/l-netperf/
[40] K. J.Duda and D.R. Cheriton. Borrowed-virtual-time (BVT) scheduling: supporting latency-sensitive threads in a general-purpose scheduler. In Proceedings of the 17th ACM SOSP, 1999.
[41] I. M.Leslie, D. Mcauley, R. Black, T. Roscoe, P.T. Barham, D. Evers, R. Fairbairns, and E. Hyden. The Design and Implementation of an Operating System to Support Distributed Multimedia Applications. IEEE Journal of Selected Areas in Communications, 1996.
[42] D. Gupta, L. Cherkasova, R. Gardner, and A. Vahdat. Enforcing Performance Isolation Across Virtual Machines in Xen. In Proceedings of the 7th International Middleware Conference, 2006.
