即時移轉是系統虛擬化中十分重要的機制，藉由預先複製的策略，以實現服務中斷時間縮短的可能性。然而，服務中斷時間會受到工作集範圍、分頁修改率，以及分頁傳送規則所影響，使得某些類型的工作可能會有較長的服務中斷時間。

本論文提出即時移轉的模型，並透過模型分析觀察到利用氣球驅動程式來限制可供虛擬機器使用的記憶體空間，可有效降低Stop-and-Copy階段的分頁傳送量，但其伴隨而來的卻是分頁替換過度而造成虛擬機器之運行效能不彰。為了讓服務中斷時間低於可容許的時間範圍，且盡可能維持虛擬機器的運行效能，本論文提出了Stop-and-Copy階段之分頁傳送量的預測流程。該流程收集了Iterative Pre-copy階段數個回合所有頁框的分頁修改狀態後，並使用兩階段式的預測規則，推測在Stop-and-Copy階段可能傳送的分頁量是否超過可容許的分頁傳送量，再根據預測結果來決定是否壓縮虛擬機器之記憶體空間，以降低Stop-and-Copy階段的分頁傳送量上限，實現服務中斷時間縮短之效果；在實機實驗中，驗證了本論文所提出之兩階段式預測規則能預測到Stop-and-Copy階段的分頁傳送量趨勢，並據以採取相對應的可使用記憶體空間壓縮，使得分頁傳送量不會超過可容許的分頁傳送量

Live migration is an important mechanism that uses the Pre-copy strategy to reduce the downtime in virtualization. However, some tasks may have a long downtime because the downtime will be affected by the sizes of the working set, modified rate, and the rule of page transference.

This thesis proposes a model of live migration, and uses the balloon driver to limit the memory space of virtual machines through the analysis of this model. It can reduce the number of pages transferred in Stop-and-Copy stage, but it also can decrease the performance of virtual machines because the pages are swapped frequently. To make the downtime being less than the acceptable interval, and keep the performance of virtual machines as much as possible. Therefore, a strategy which predicts the number of pages transferred in Stop-and-Copy stage is proposed in this thesis. The strategy collects the states of all frames at many iterations in Iterative Pre-copy stage, and then uses the two-phase predicting rule to predict the number of pages transferred possibly in Stop-and-Copy stage whether that is greater than the number of acceptable pages. According to the result of prediction, the strategy decides whether to compress the memory space of virtual machines for decreasing the upper bound of the number of pages transferred in Stop-and-Copy stage, achieves the result that reduces the downtime. It has been shown that the strategy can predict the trend of the number of pages transferred in Stop-and-Copy stage, and then select the corresponding action that compress the available memory, make the number of page transferred would be less than the number of acceptable pages

摘要 i
Abstract ii
誌謝 iii
目錄 iv
圖目錄 v
表目錄 vi
1 緒論 1
1.1 背景知識 1
1.2 研究動機與目的 8
1.3 論文架構 8
2 文獻回顧與探討 10
3 Pre-copy策略分析 19
3.1 通用模型 19
3.2 效能模型 22
3.3 Xen的即時移轉 24
3.3.1 即時移轉流程與分頁傳送規則 25
3.3.2 通用模型與效能模型 31
3.3.3 分頁替換 39
3.3.4 駐留集與工作集 42
3.3.5 記憶體壓縮與運行效能 46
4 Stop-and-Copy階段之分頁傳送量預測 52
4.1 終止條件探討與修改 52
4.2 兩階段式的預測規則 56
4.3 預測流程 62
5 實機實驗 66
5.1 實驗環境 66
5.2 實驗結果分析 67
5.2.1 最後回合分頁傳送量 67
5.2.2 運行效能 82
6 結論與未來研究方向 86
Reference 88

[1] salesforce.com, http://www.salesforce.com/tw/

[2] Peter Mell and Timothy Grance, "The NIST Definition of Cloud Computing," Sep. 2011, http://csrc.nist.gov/publications/nistpubs/800-145/SP800-145.pdf

[3] Xin Zhao, Kevin Borders, Atul Prakash, "Virtual Machine Security Systems," Book Chapter in ADVANCES IN COMPUTER SCIENCE AND ENGINEERING, pp.339-365, 2009.

[4] 孫昱, 虛擬機Xen及其實時遷移技術研究, 上海交通大學碩士論文, 2008.

[5] VMware.com, "Understanding Full Virtualization, Paravirtualization, and Hardware Assist", 2007, http://www.vmware.com/files/pdf/VMware_paravirtualization.pdf

[6] 陳瀅, 雲端策略：雲端運算與虛擬化技術, 天下雜誌, 2010.

[7] Paul R. Barham, Boris Dragovic, Keir A. Fraser, Steven M. Hand, Timothy L. Harris, Alex C. Ho, Evangelos Kotsovinos, Anil V.S. Madhavapeddy, Rolf Neugebauer, Ian A. Pratt, and Andrew K. Warfield, "Xen 2002," 2003, http://www.cl.cam.ac.uk/techreports/UCAM-CL-TR-553.pdf

[8] 辛曉慧,"Xen內存虛擬化實現," 開放系統世界, 2006, http://tcloud.sjtu.edu.cn/wiki/images/9/9b/Xen硬件虚拟机的内存虚拟化实现.pdf

[9] XCP FAQ Dynamic Memory Control, http://wiki.xen.org/wiki/XCP/FAQ_Dynamic_Memory_Control

[10] Christopher Clark, Keir Fraser, Steven Hand, Jacob Gorm Hansen, Eric Jul, Christian Limpach, Ian Pratt and Andrew Warfield, "Live Migration of Virtual Machines," NSDI'05 Proceedings of the 2nd conference on Symposium on Networked Systems Design & Implementation, Vol.2, pp.273-286, 2005.

[11] Marvin M. Theimer, Keith A. Lantz and David R. Cheriton, "Preemptable remote execution facilities for the V-system," SOSP '85 Proceedings of the tenth ACM symposium on Operating systems principles, pp.2-12, 1985.

[12] Michael R. Hines and Kartil Gopalan, "Post-copy based live virtual machine migration using adaptive pre-paging and dynamic self-ballooning," VEE'09 Proceedings of the 2009 ACM SIGPLAN/SIGOPS international conference on Virtual execution environments, pp.51-60, 2009.

[13] Anja Strunk, "Costs of Virtual Machine Live Migration: A Survey," 2012 IEEE Eighth World Congress on Services, pp.323-329, 2012.

[14] Felix Salfner, Peter Tröger and Andreas Polze, "Downtime Analysis of Virtual Machine Live Migration," The Fourth International Conference on Dependability(DEPEND 2011), pp.100-105, 2011.

[15] Fei Ma, Feng Liu and Zhen Liu, "Live Virtual Machine Migration based on Improved Pre-copy Approach," 2010 IEEE International Conference on Software Engineering and Service Sciences (ICSESS), pp.230-233, 2010.

[16] Bolin Hu, Zhou Lei, Yu Lei, Dong Xu and Jiandum Li, "A Time-Series Based Precopy Approach for Live Migration of Virtual Machines," 2011 IEEE 17th International Conference on Parallel and Distributed Systems (ICPADS), pp.947-952, 2011.

[17] Cho-Chin Lin, Yu-Chi Huang and Zong-De Jian, "A Two-phase Iterative Pre-copy Strategy for Live Migration of Virtual Machines," 2012 8th International Conference on Computing Technology and Information Management (ICCM), Vol.1, pp.29-34, 2012.

[18] SUN Guo-fei, GU Jian-hua, HU Jin-hua and ZHAO Tian-hai, "Improvement of Live Memory Migration Mechanism for Virtual Machine Based on Pre-copy," Computer Engineering, Vol.37, No.13, 2011.

[19] Yanqing Ma, Hongbo Wang, Jiankang Dongm Yangyang Li and Shiduan Cheng, "ME2: Efficient Live Migration of Virtual Machine with Memory Exploration and Encoding," 2012 IEEE International Conference on Cluster Computing (CLUSTER), pp.610-613. 2012.

[20] Sherif Akoush, Ripduman Sohan, Andrew Rice, Andrew W. Moore and Andy Hopper, "Predicting the Performance of Virtual Machine Migration,", 2010 IEEE International Symposium on Modeling, Analysis & Simulation of Computer and Telecommunication Systems (MASCOTS), pp.37-46, 2010.