A reuse distance based precopy approach to improve live migration of virtual machines

Virtualization enables data centers to share physical resources between virtual private servers (VPS) cost-effectively. Live migration of virtual machines (VM) is one of the most important capabilities that have been provided by the virtualization. It enables the data center's administrators to apply the load balancing properly. Sending large number of VM's memory pages wastes the data center's bandwidth. It also decreases the quality of service during live migration. By reducing the number of sent pages, live migration time is reduced. Using an effective approach for sending VM's pages leads to shorter downtime and migration time compared to pre-copy. This paper proposes an algorithm for tracking updated memory pages and making decision for transferring dirty pages during each iterated migration phase. In this paper, the concept of reuse distance is adopted to reduce the transferring number of dirtied pages. According to the experimental results, when the system is under heavy workload or running applications inside the VM are write-intensive; the reuse-distance approach has better performance than traditional pre-copy in terms of the number of transferred pages, total migration time and downtime. Experiments show that with increasing workload, reduction in the number of sent pages reaches 68% without increasing systems' overhead during live migration.

[1]  Keqiu Li,et al.  Hierarchical Copy Algorithm for Xen Live Migration , 2010, 2010 International Conference on Cyber-Enabled Distributed Computing and Knowledge Discovery.

[2]  S. Hand,et al.  Xen 3.0 and the Art of Virtualization , 2010 .

[3]  Kartik Gopalan,et al.  Post-copy based live virtual machine migration using adaptive pre-paging and dynamic self-ballooning , 2009, VEE '09.

[4]  Yutao Zhong,et al.  Predicting whole-program locality through reuse distance analysis , 2003, PLDI.

[5]  Guan Haibing Live Migration of Virtual Machines , 2008 .

[6]  Hai Jin,et al.  Live virtual machine migration with adaptive, memory compression , 2009, 2009 IEEE International Conference on Cluster Computing and Workshops.

[7]  Umesh Deshpande,et al.  Post-copy live migration of virtual machines , 2009, OPSR.

[8]  Eli M. Dow,et al.  Running Xen: A Hands-On Guide to the Art of Virtualization , 2008 .

[9]  Meina Song,et al.  Live memory migration with matrix bitmap algorithm , 2010, 2010 IEEE 2nd Symposium on Web Society.

[10]  Yaozu Dong Extending Xen* with IntelŴVirtualization Technology , 2006 .

[11]  Yaozu Dong,et al.  Extending Xen* with Intel Virtualization Technology , 2006 .

[12]  Gil Neiger,et al.  Intel ® Virtualization Technology for Directed I/O , 2006 .

[13]  Arun Venkataramani,et al.  Black-box and Gray-box Strategies for Virtual Machine Migration , 2007, NSDI.

[14]  Song Jiang,et al.  CLOCK-Pro: An Effective Improvement of the CLOCK Replacement , 2005, USENIX Annual Technical Conference, General Track.

[15]  Carl A. Waldspurger,et al.  Memory resource management in VMware ESX server , 2002, OSDI '02.