臺灣博碩士論文加值系統

虛擬機器自動容錯技術是採用備援虛擬機器的方式，將主要虛擬機器的狀態不斷的同步至另一台實體機器，當主要虛擬機器因任何因素導致服務終止時，備援虛擬機器會立刻接手繼續提供服務，讓主要虛擬機器原先運行的服務不會因此中斷。但根據我們的研究，目前基於KVM的開源容錯如Kemari和Micro-Checkpointing並無法提供一個可以讓網路服務順暢運作的環境，因此中央大學資訊工程研究所的研究團隊提出了一個針對網路服務的容錯架構M-FTVM，而此架構能降低容錯系統對於雲端網路服務的影響。本研究承襲了這樣的容錯架構，並進行錯誤修正，解決了前版本備援虛擬機器對主要虛擬機器生命週期產生影響的問題，經過實驗的測試，問題修正後效能與前版本並無太大損失，而且和其他基於KVM的開源容錯相比我們的容錯系統在網路服務方面效能優異許多。

Automatic fault tolerance for virtual machine usually uses a backup virtual machine to protect the primary virtual machine, which provides services to end users. The primary and the backup virtual machines have to run on different physical machines and keep their states synchronized. As the primary virtual machine fails, the backup virtual machine has to take the control and keeps the services on the virtual machine alive. Based on our study, the existing open-source projects of virtual machine fault tolerance, Kemari and Micro-Checkpointing, do not have good performance when hosting a network service. Therefore, we have proposed a novel design of a fault-tolerant virtual machine based on KVM, namely M –FTVM. We have also implemented a prototype of the proposed fault-tolerant virtual machine, and keep working on improving its performance and correctness. This paper focus on the issue of the virtual machine life cycle on the original M-FTVM. That is, the backup virtual machine may cause the primary virtual machine unstable and affects the life cycle of virtual machine. Therefore, we propose a new design for the virtual machine life cycle of M-FTVM, and modified the implementation of M-FTVM accordingly. The experiment results show that, the new design does not cause much overhead when compared with the original M-FTVM.

摘要 i
Abstract ii
圖目錄 v
表目錄 vii
第一章 緒論 1
1-1 研究背景 1
1-2 研究動機與貢獻 3
1-3 論文架構 4
第二章 相關研究 6
2-1 背景知識 6
2-1-1 KVM(Kernel-based Virtual Machine)&QEMU 6
2-2 主要虛擬機容錯策略 6
2-3 KVM容錯系統 7
2-3-1 Kemari 7
2-3-2 Micro-Checkpointing 9
2-4 VMware高可用性服務架構 10
第三章 系統架構 11
3-1 主要架構 11
3-2 系統運作流程 12
3-3 Correctness 15
3-4 Self-checkpoint &Active-checkpoint 19
第四章 問題修正 22
4-1 VM Life Cycle 22
4-2 問題修正 24
第五章 實驗結果 28
5-1 實驗環境與架構 28
5-2 DVD Store Benchmark實驗結果與分析 31
5-2-1 Operation per Minute 31
5-2-2 Response Time 32
5-2-3 每執行一個operation所消耗的頻寬 34
5-3 Yahoo! Cloud Serving Benchmark實驗結果與分析 35
5-3-1 YCSB實驗數據 36
第六章 結論及未來研究方向 40
第七章 參考文獻 42

[1] G. J. Popek and R. P. Goldberg, "Formal requirements for virtualizable third generation architectures," Commun. ACM, vol. 17, pp. 412-421, 1974.
[2] R. P. Goldberg, "Survey of virtual machine research," Computer, vol. 7, pp. 34-45, 1974 1974.
[3] Ponemon. (2016). Cost of Data Center Outages. Available: http://datacenterfrontier.com/cost-of-data-center-outages/
[4] J. Gray and D. P. Siewiorek, "High-availability computer systems," Computer, vol. 24, pp. 39-48, 1991.
[5] D. J. Scales, M. Nelson, and G. Venkitachalam, "The design and evaluation of a practical system for fault-tolerant virtual machines," Technical Report VMWare-RT-2010-001, VMWare2010.
[6] C.-H. Chen, "基於KVM的網路服務高可靠性容錯同步架構," Master's thesis, National Central University, 2014.
[7] KVM – Kernel-based Virtual Machine | Red Hat [Online]. Available: https://www.linux-kvm.org
[8] M. Zabaljauregui, "Hardware assisted virtualization intel virtualization technology," 2008.
[9] AMD. (2007). Putting Server Virtualization to Work. Available: https://www.redhat.com/f/pdf/virtualization/amd_Virtualization_WP.pdf
[10] F. Bellard, "QEMU, a fast and portable dynamic translator," in USENIX Annual Technical Conference, FREENIX Track, 2005, pp. 41-46.
[11] B. Cully, G. Lefebvre, D. Meyer, M. Feeley, N. Hutchinson, and A. Warfield, "Remus: High availability via asynchronous virtual machine replication," in Proceedings of the 5th USENIX Symposium on Networked Systems Design and Implementation, 2008, pp. 161-174.
[12] T. C. Bressoud and F. B. Schneider, "Hypervisor-based fault tolerance," ACM Transactions on Computer Systems (TOCS), vol. 14, pp. 80-107, 1996.
[13] Y. Tamura, K. Sato, S. Kihara, and S. Moriai, "Kemari: Virtual machine synchronization for fault tolerance," 2008.
[14] Features/MicroCheckpointing - QEMU. Available: http://wiki.qemu.org/Features/MicroCheckpointing
[15] M. Xu, V. Malyugin, J. Sheldon, G. Venkitachalam, and B. Weissman, "Retrace: Collecting execution trace with virtual machine deterministic replay," in In Proceedings of the 3rd Annual Workshop on Modeling, Benchmarking and Simulation, MoBS, 2007.
[16] R. Jhawar, V. Piuri, and M. Santambrogio, "Fault tolerance management in cloud computing: A system-level perspective," IEEE Systems Journal, vol. 7, pp. 288-297, 2013.
[17] Dvd store. Available: http://www.delltechcenter.com/page/DVD+store
[18] Yahoo! Cloud Serving Benchmark Available: https://github.com/brianfrankcooper/YCSB/wiki