跳到主要內容

簡易檢索 / 詳目顯示

研究生: 林奇賦
Chi-fu Lin
論文名稱: 使用LRO於虛擬機分享高速I/O裝置之效能增進
Using LRO to Improve High Speed I/O on Virtual Machine Environments
指導教授: 曾黎明
Li-Ming Tseng
口試委員:
學位類別: 碩士
Master
系所名稱: 資訊電機學院 - 資訊工程學系
Department of Computer Science & Information Engineering
畢業學年度: 97
語文別: 中文
論文頁數: 40
中文關鍵詞: LRO虛擬機器Giga網路
外文關鍵詞: Virtual machine, Gigabit Ethernet, LRO
相關次數: 點閱:13下載:0
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報
  • 隨著電腦軟硬體能力的大幅提昇,虛擬機器技術的應用也日趨普遍,虛擬機的技術可以依使用者的需要用來解決許多系統方面的問題,例如容錯(fault tolerance)、系統搬移(migration)、入侵偵測(intrusion detection)或偵錯(debugging)等。除了指令虛擬化之外,I/O介面必須在軟硬體溝通的程序中加入轉換的機制,由於周邊裝置種類繁多,所以也有很多不同的技術。目前在VMM中提供的I/O裝置虛擬化模型,常是系統整體效能無法上升的一大瓶頸,而我們發現很多研究藉由避開VMM對I/O的不必要監控,可以得到效能的提升。
    目前Gigabit(1000M)網路已經普及化,高速網路傳輸的效能亦需要重視及優化,而本論文將針對TCP的接收,實做LRO(Large Receive Offload)於虛擬機器中,並針對虛擬機器傳遞封包的機制以不同的架構傳遞,分析效能上的改進與否。


    With the increasing of modern computing power that greatly enhance the feasibility of virtual machine technology. VM technology is usually used to solve many management problems, such as fault-tolerant, process migration, intrusion detection or system debugging, and so on. In the virtual machine architecture, VM Monitor (VMM) is the key component for virtualization. Except to the virtualization of instruction set and memory space, the complexity of I/O devices and software drivers caused the raising need for I/O virtualization. Communication compatibility and performance of I/O software to the real hardware devices are the major concern.
    Gigabit Ethernet NICs are increasingly common. It is very important to focus on performance. In this paper, we implemented Large Receive Offload (LRO) in virtual network driver. Finally we will analysis modified driver and measure its effectiveness.

    摘要 i Abstract ii 誌謝 iii 目錄 iv 圖目錄 v 表目錄 vi 第一章 緒論 1 第二章 相關研究 3 2.1虛擬機器平台 3 2.2 I/O虛擬化技術 (I/O Virtualization) 4 2.3 XEN網路傳輸架構 (Networking in Xen) 8 2.4 KVM網路傳輸架構 (Networking in KVM) 9 第三章 系統設計 11 3.1 LRO(Large Receive Offload) 11 3.2效能瓶頸分析 13 3.2.1 TCP Protocol性能瓶頸 13 3.2.2中斷Overhead分析 14 3.3系統分析 16 3.3.1實體網卡驅動分析 16 3.3.2網路虛擬化VNIF(Virtual Network Interface) 20 3.3.3 Xen中斷分析 24 3.4系統架構 27 3.4.1實體網卡驅動修改 27 3.4.2虛擬網卡驅動修改 28 第四章 效能評估 30 4.1實驗環境 30 4.2 CPU Loading量測 30 4.3 Throughput量測 33 4.4 Interrupt量測 35 第五章 討論與未來展望 36 第六章 結論 37 參考文獻 38 附錄:研究論文計畫口試建議改進事項 40

    [1]The Internet Suspend/Resume (ISR) project, http://isr.cmu.edu/.
    [2]Revirt, http://www.eecs.umich.edu/CoVirt/.
    [3]VMware, URL http://www.VMware.com.
    [4]Xen, URL http://www.xen.org/.
    [5]J. Liu, W. Huang, B. Abali, D. K. Panda. High Performance VMM-Bypass I/O in Virtual Machines, USENIX Annual Technical Conference (USENIX’06), May, 2006
    [6]L. Grossman, “Large Receive Offload implementation in Neterion 10GbE Ethernet driver,” Proc. Linux Symp., vol. One, pp.195-200, July 2005.
    [7]A. Gallatin, “Re: [RFC 0/1] lro: Generic Large Receive Offload for TCP traffic,” July 2007, http://lkml.org/lkml/2007/7/25/313.
    [8]QEMU, http://fabrice.bellard.free.fr/qemu/.
    [9]Fabrice Bellard, 「QEMU, a Fast and Portable Dynamic Translator,」 FREENIX Track: USENIX Annual Technical Conference, 2005.
    [10]Himanshu Raj, Karsten Schwan. 「Implementing a Scalable Self-Virtualizing Network Interface on an Embedded Multicore Platform」, CERCS, Georgia Institute of Technology Atlanta, 30332.
    [11]P. Willmann, J. Shafer, D. Carr, A. Menon, S. Rixner, A. L. Cox and W. Zwaenepoel“Concurrent Direct Network Access for Virtual Machine Monitors”, High Performance Computer Architecture, HPCA, IEEE 13th International Symposium on, pp. 306-317, Feb.2007
    [12]Dong-Jae Kang, Chei-Yol Kim, Kang-Ho Kim, Sung-In Jung, "Proportional Disk I/O Bandwidth Management for Server Virtualization Environment," iccsit,pp.647-653, 2008 International Conference on Computer Science and Information Technology, 2008
    [13]Weikuan Yu, Jeffrey S. Vetter: Xen-Based HPC: A Parallel I/O Perspective. CCGRID 2008: 154-161
    [14]Abramson, D., Jackson, J., Muthrasanallur, S., Neiger, G., Regnier, G., Sankaran, R., Schoinas, I., Uhlig, R., Vembu, B., and Wiegert, J. “Intel virtualization technology for directed I/O”. Intel Technology Journal 10, 3 (August 2006). http://www.intel.com/technology/itj/2006/v10i3/
    [15]AMD, http://www.amd.com/us-en/0,,3715_15781,00.html?redir=SWOP08
    [16]J.-B. Themann, “Re: [RFC 0/1] lro: Generic Large Receive Offload for TCP traffic,” July 2007, http://lkml.org/lkml/2007/7/20/250.
    [17]Qiang L, Qinfen Hao, Limin Xiao, Zhoujun Li, "VM-based Architecture for Network Monitoring and Analysis," Young Computer Scientists, 2008. ICYCS 2008. The 9th International Conference for , vol., no., pp.1395-1400, 18-21 Nov. 2008
    [18]Intel Corp. Intel 82598 10 GbE ethernet controller open source datasheet, 2008. Revision 2.5.
    [19]Ram, K. K., Santos, J. R., Turner, Y., Cox, A. L., and Rixner, S. 2009. Achieving 10 Gb/s using safe and transparent network interface virtualization. In Proceedings of the 2009 ACM SIGPLAN/SIGOPS international Conference on Virtual Execution Environments (Washington, DC, USA, March 11 - 13, 2009). VEE ''09. ACM, New York, NY, 61-70.
    [20]Netperf: http://www.netperf.org/netperf/
    [21]David Chisnall. The Definite Guide to the Xen Hypervisor. Prentic Hall: 2007.
    [22]Intel PCIe* GbE Controllers Open Source Software Developer’s Manual http://download.intel.com/design/network/manuals/31608004.pdf
    [23]Sheng Wang, Jin-shu Su, " Design and Implementation of FPGA-based TOE System", Computer science 2008, China.

    QR CODE
    :::