隨著以雲端服務為導向的商業模式蓬勃發展和基礎架構的雲端服務(IaaS)的日益成熟，過去傳統方式的建置服務已無法滿足使用者的需求，許多企業或是使用者選擇將服務放置在IaaS雲端服務供應商，IaaS是由實體和虛擬資源組成，其中虛擬化技術可以透過程式管理提升硬體使用率，搭配動態遷移技術將虛擬機從原本的實體機轉移到另一台實體機，使用者透過動態調整虛擬機位置，來確保最大限度的減少服務停機時間，同時減少中間消耗的能源。本論文以OpenStack的雲端平台系統做為虛擬機動態遷移的基礎，實驗虛擬機在實體機器上進行的撤離時間預測的準確性，並在其中比較不同因素對於撤離時間預測的影響並分析其造成的原因。

With business models for cloud computing flourishing and the maturity of Infrastructure as a service(IaaS), the traditional ways of building services in the past can no longer satisfy the demand of users, accordingly, many companies and users prefer to place their services on IaaS cloud servers.
IaaS is composed of physical and virtual resources, its virtualization can increase hardware usage rate through program management, and it uses evacuate to move a running virtual machine between different physical machines. Users can dynamically adjust the position of the virtual machine to ensure that the service downtime is minimized and reduces energy consumption during the process at the same time.
The thesis uses the cloud platform system of OpenStack as the basis for the evacuate of the virtual machine to test the accuracy of the virtual machine's pre-evacuation time from the physical machine and comparing the influence of different factors on the pre-evacuation time as well as cause analysis.

[1] “Cloud Market Share – a Look at the Cloud Ecosystem in 2021” Edward Jones, April 21, 2021.Available:https://kinsta.com/blog/cloud-market-share/#an-overview-of-the-cloud-computing-market-in-2021
[2] “aws service level agreements (SLAs)” Available:https://aws.amazon.com/tw/legal/service-level-agreements/
[3] “How vSphere HA Works” Available:https://docs.vmware.com/en/VMware-vSphere/7.0/com.vmware.vsphere.avail.doc/GUID-33A65FF7-DA22-4DC5-8B18-5A7F97CCA536.html
[4] T. L. Nguyen and A. Lebre, “Virtual Machine Boot Time Model,” in 2017 25th Euromicro International Conference on Parallel, Distributed and Network-based Processing (PDP), St. Petersburg, Russia, 2017, pp. 430–437, doi: 10.1109/PDP.2017.58.
[5] 黃嘉俊。「精準虛擬機開機時間預測與在虛擬機快速撤離上的應用」。碩士論文，國立中央大學資訊工程學系，2021。
[6] R. Sharma, Learning OpenStack High Availability. Packt Publishing Ltd, 2015.
[7] “Messaging that just works — RabbitMQ.” Available:https://www.rabbitmq.com/ (accessed Jul. 12, 2020).
[8]“libvirt: The virtualization API.”
Available:https://libvirt.org/ (accessed Jul. 12, 2020).
[9] Understanding Request-Process Flows in OpenStack for Troubleshooting - Glenn West. 2014.
[10] ”OpenStack Docs: Filter Scheduler.” Available:https://docs.openstack.org/nova/latest/user/filter-scheduler.html (accessed Jul. 13, 2020).
[11] “ProximityScheduler - OpenStack.” Available:https://wiki.openstack.org/wiki/ProximityScheduler (accessed Nov. 11, 2020)
[12] J. Hamilton, “Architecture for Modular Data Centers,” Jan. 2007.
[13]“Amazon Elastic Block Store (EBS) - Amazon Web Services.” Available:https://aws.amazon.com/ebs/ (accessed Jul. 25, 2020).
[14]“Persistent Disk,” Google Cloud. Available:https://cloud.google.com/persistent-disk (accessed Jul. 25, 2020).
[15] H. Lu, B. Saltaformaggio, C. Xu, U. Bellur, and D. Xu, “BASS: Improving I/O Performance for Cloud Block Storage via Byte-Addressable Storage Stack,” Oct. 2016, pp. 169–181, doi: 10.1145/2987550.2987557.
[16] J. Satran, K. Meth, C. Sapuntzakis, M. Chadalapaka, and E. Zeidner, “Internet Small Computer Systems Interface (iSCSI),” RFC Editor, RFC3720, Apr. 2004. doi: 10.17487/rfc3720.
[17]“A Performance Comparison of NFS and iSCSI for IP-Networked Storage.” Available:https://www.usenix.org/legacy/publications/library/proceedings/fast04/tech/full_papers/radkov/radkov_html/head.html (accessed Jul. 26, 2020).
[18]“Storage Area Network (SAN) Market Analysis and Review 2019 - 2029 | Future Market Insights (FMI).” Available:https://www.futuremarketinsights.com/reports/storage-area-network-san-market (accessed Jul. 26, 2020).
[19]“TPC.” Available:http://www.tpc.org/tpch/ (accessed Jul. 26, 2020).
[20]“Ubuntu Manpage: iperf - perform network throughput tests.” Available:http://manpages.ubuntu.com/manpages/xenial/en/man1/iperf.1.html (accessed Jul. 10, 2020).
[21]“LINPACK.” Available:http://www.netlib.org/linpack/ (accessed Jul. 10, 2020).
[22]“Cachebench Home Page.”
Available:http://icl.cs.utk.edu/llcbench/cachebench.html (accessed Jul. 10, 2020).
[23]“Ubuntu Manpage: stress - tool to impose load on and stress test systems.” Available:http://manpages.ubuntu.com/manpages/xenial/man1/stress.1.html (accessed Jul. 10, 2020).
[24] J. T. Piao and J. Yan, “A Network-aware Virtual Machine Placement and Migration Approach in Cloud Computing,” In 2010 Ninth International Conference on Grid and Cloud Computing, Nov. 2010, pp. 87–92, doi: 10.1109/GCC.2010.29.
[25]“Amazon Elastic Block Store (Amazon EBS) - Amazon Elastic Compute Cloud.” Available:https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/AmazonEBS.html (accessed Jul. 11, 2020).
[26] H. Wu et al. ” A Reference Model for Virtual Machine Launching Overhead ” IEEE Transactions on Cloud Computing, vol. 4, no. 3, pp. 250–264, Jul. 2016, doi: 10.1109/TCC.2014.2369439.
[27] M. Mao and M. Humphrey. “A performance study on the vm startup time in the cloud.” In Cloud Computing (CLOUD), 2012 IEEE 5th International Conference on, pages 423–430. IEEE, 2012.