期刊名称：International Journal of Innovative Research in Science, Engineering and Technology
印刷版ISSN：2347-6710
电子版ISSN：2319-8753
出版年度：2018
卷号：7
期号：3
页码：2793
DOI：10.15680/IJIRSET.2018.0703136
出版社：S&S Publications
摘要：In cloud computing Virtualization is a key technology that allows several operating system instances torun on the same host machine, to enhance resource utilization and to improve the application response time.Virtualization needs a software layer called as hypervisor, on top of the hardware layer. Virtual Machines are run ontop of the hypervisor, which provides hardware resources to the VMs (Guest Machines). Virtualization enablescomplete utilization of a physical machine in a cost-effective manner. However, varying and continuous load frommultiple users may overload a physical machine and creates overhead on performance, reliability and other servicelevel-agreement (SLA) parameters. In addition to enabling higher utilization of hardware resources, migration of VMsfrom one host to another in same DC or another DC is needed. Past research works confirmed that two importantparameters affecting live migration are downtime and total migration time. Traditional pre-copy and post-copyapproaches can reduce the downtime but the total migration time is too long due to repeated retransmissions. Therefore,to improve the service downtime and total migration time and to migrate multiple VMs An optimal technique wasproposed. This paper contains analysis of performance metrics like application service downtime, total migration timeand amount of data transferred. CPU, memory and storage data is transferred during the process of VM migration andwe identify the category of data that needs to be transferred in each stage.
关键词：Cloud Computing; Virtualization; VM Live Migration; Pre-copy; Post-copy; Optimal Technique;Performance Metrics; Multiple VMs Live Migration.
