A quantitative study of virtual machine live migration
Proceedings of the 2013 ACM Cloud and Autonomic Computing Conference
| Hi-index | 0.00 |
Live migration of virtual machine plays an important role in data center, which can successfully migrate virtual machine from one physical machine to another with only slight influence on upper workload. It can be used to facilitate hardware maintenance, load balancing, fault-tolerance and power-saving, especially in cloud computing data centers. Although the pre-copy is the prevailing approach, it cannot distinguish which memory page is used, resulting in transferring large amounts of useless memory pages. This paper presents a novel approach Memory Exploration and Encoding (ME2), which first identifies useful pages and then utilizes Run Length Encode algorithm to quickly encode memory, to efficiently decrease the total transferred data, total migration time and downtime. Experiments demonstrate that ME2 can significantly decrease 50.5% of total transferred data, 48.2% of total time and 47.6% of downtime on average compared with Xen's pre-copy algorithm.