The Research on Live Migration Strategy of OpenStack Virtual Machine based on NFS Shared Storage

• Main Authors: Wu, Hsin-Ting, 吳馨婷
• Other Authors: Tsai, Kuo-Hui
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/2bqvjt

碩士 === 國立臺灣海洋大學 === 資訊工程學系 === 106 === With the rapid development of information technology, the advancement of cloud technology makes the network infrastructure popular. To reduce the cost of software licensing, either government, enterprise, or individual can now build their own cloud service by using open source technology. In the cloud infrastructure, migration technology has become an essential function of public cloud and private cloud. The live migration technique helps the maintenance of the virtual machine, such that the overloading problem can be efficiently solved without service interruption. In this study, an open source OpenStack was used to build a cloud service platform, and NFS (Network File System) shared storage was used to implement live migration of virtual machines. Image files of virtual machines were placed in directories shared by source and target hosts. The virtual machine migrates the vCPU execution status and memory increment data to the target host through the network. Shell scripts are designed to generate the network traffic and to collect migration statistics. In coordination with Ceilometer, a monitoring module in OpenStack, the user is prompted to migrate the virtual machine to some light-loaded machine. To examine the quality of the migration, web server stress tool is employed to generate HTTP requests. This study shows that the migration timing algorithm enhances the overall system performance as well as service quality.