基于副本的跨数据中心虚拟机快速迁移算法

删除或更新信息，请邮件至freekaoyan#163.com(#换成@)

清华大学辅仁网/2017-07-07

基于副本的跨数据中心虚拟机快速迁移算法

刘圣卓^1,2, 姜进磊^1,2, 杨广文^1,2

1. 清华大学计算机科学与技术系, 北京 100084;
2. 清华信息科学与技术国家实验室(筹), 北京 100084

Fast cross-datacenter virtual machine migration algorithm based on duplication

LIU Shengzhuo^1,2, JIANG Jinlei^1,2, YANG Guangwen^1,2

1. Department of Computer Science and Technology, Tsinghua University, Beijing 100084, China;
2. Tsinghua National Laboratory for Information Science and Technology, Tsinghua University, Beijing 100084, China

摘要:

输出: BibTeX | EndNote (RIS)

摘要跨广域网的虚拟机动态迁移是多数据中心云计算环境的重要技术支撑。与局域网上相比,跨广域网的动态迁移由于受到带宽的限制而面临更多的技术挑战。基于跨数据中心虚拟机迁移的特性,提出了多数据中心虚拟机快速动态迁移架构和跨数据中心快速动态镜像迁移(FlimCD)算法。FlimCD算法利用脏块去冗、缺块预取和热区排序传输等优化方法,减少了虚拟机整体迁移时间,并降低了迁移对性能的影响。实验表明: 对于中等I/O强度工作流,FlimCD算法的迁移时间比前拷贝方法的减少64.3%;FlimCD的优化机制使后拷贝阶段的缺块率下降49%以上。

关键词 ：虚拟机,广域网迁移,动态存储迁移,去冗余,多数据中心

Abstract：Live migration of virtual machines (VM) over wide-area networks (WAN) is key to cloud computing operations in a multi-datacenter environment. Compared with the migration over a local-area network (LAN), the live migration over a WAN faces more technical challenges due to the limited bandwidth. A live cross-datacenter VM migration infrastructure was developed in a fast live image migration cross-datacenters (FlimCD) method, which makes full use of dirty blocks deduplication, block pre-fetching and hot-area scheduled transmissions to reduce the overall migration time and the impact of the VM migration on the performance. Tests show that, for moderately I/O intensive workloads, the migration time is reduced by 64.3% compared with that of the pre-copy algorithm. The current optimization mechanisms reduce the block miss rate by 49% in the post-copy phase.

Key words：virtual machine wide-area migration live storage migration deduplication multi-datacenter

收稿日期: 2015-02-16 出版日期: 2015-08-04

ZTFLH:

TP316.1

通讯作者:姜进磊,副教授,E-mail:jjlei@tsinghua.edu.cnE-mail: jjlei@tsinghua.edu.cn

引用本文:

刘圣卓, 姜进磊, 杨广文. 基于副本的跨数据中心虚拟机快速迁移算法[J]. 清华大学学报（自然科学版）, 2015, 55(5): 579-584.
LIU Shengzhuo, JIANG Jinlei, YANG Guangwen. Fast cross-datacenter virtual machine migration algorithm based on duplication. Journal of Tsinghua University(Science and Technology), 2015, 55(5): 579-584.

链接本文:

http://jst.tsinghuajournals.com/CN/或 http://jst.tsinghuajournals.com/CN/Y2015/V55/I5/579

图表:

图１　跨数据中心虚拟机快速动态迁移架构

图２　FlimCD 算法时序图

表１　虚拟机镜像的分布

图３　全镜像迁移和部分迁移比较

图４　迁移时间

图５　迁移过程的数据传输量

图６　迁移过程的缺块率

参考文献:

[1]	Mahalingam M, Dutt D, Duda K, et al. VXLAN: A Framework for Overlaying Virtualized Layer 2 Networks over Layer 3 Networks [S/OL]. [2015-01-25]. https://tools.ietf.org/html/draft-mahalingam-dutt-dcops-vxlan-00.
[2]	Farinacci D, Fuller V, Meyer D, et al. The Locator/ID Separation Protocol (LISP) [S/OL]. [2015-01-25]. http://tools.ietf.org/html/rfc6830.
[3]	Clark C, Fraser K, Hand S, et al. Live migration of virtual machines [C]//Proc of NSDI 2005. Berkeley, CA, USA: USENIX Association, 2005: 273-286.
[4]	Nelson M, Lim B H, Hutchins G. Fast transparent migration for virtual machines [C]//Proceedings of the Annual Conference on USENIX Annual Technical Conference. Berkeley, CA, USA: USENIX Association, 2005: 391-394.
[5]	Hines M R, Deshpande U, Gopalan K. Post-copy live migration of virtual machines [C]//Proc of SIGOPS'09. Montana, USA: Association for Computing Machinery Special Interest Group on Operating Systems, 2009: 14-26.
[6]	Bradford R, Kotsovinos E, Feldmann A, et al. Live wide-area migration of virtual machines including local persistent state [C]//Proc of VEE'07. New York, NY, USA: ACM, 2007: 169-179.
[7]	Mashtizadeh A, Celebi E, Garfinkel T, et al. The design and evolution of live storage migration in VMware ESX [C]//Proceedings of the 2011 USENIX Conference on USENIX Annual Technical Conference. Berkeley, CA, USA: USENIX Association, 2011: 14-14.
[8]	Hirofuchi T, Nakada H, Ogawa H, et al. A live storage migration mechanism over wan and its performance evaluation [C]//Proc of VTDC'09. New York, NY, USA: ACM, 2009: 67-74.
[9]	Hirofuchi T, Ogawa H, Nakada H, et al. A live storage migration mechanism over WAN for relocatable virtual machine services on clouds [C]//Proceedings of the 9th IEEE/ACM International Symposium on Cluster Computing and the Grid, 2009. Washington DC, USA: IEEE Computer Society, 2009: 460-465.
[10]	Luo Y, Zhang B, Wang X, et al., Live and incremental whole-system migration of virtual machines using block-bitmap [C]//2008 IEEE International Conference on Cluster Computing. Tsukuba, Japan: IEEE Computer Society, 2008: 99-106.
[11]	Zheng J, Ng T S E, Sripanidkulchai K. Workload-aware live storage migration for clouds [C]//Proceedings of the 7th ACM SIGPLAN/SIGOPS International Conference on Virtual Execution Environments. New York, NY, USA: ACM, 2011: 133-144.
[12]	FIU. FIU IODedup mail server trace [Z/OL]. [2015-01-05]. http://iotta.snia.org/traces/391.
[13]	Takahashi K, Sasada K, Hirofuchi T. A fast virtual machine storage migration technique using data deduplication [C]//Proc of CLOUD COMPUTING 2012. Nice, France: IARIA, 2012: 57-64.
[14]	Zhao X, Zhang Y, Wu Y, et al. Liquid: A scalable deduplication file system for virtual machine images [J]. IEEE Trans Parallel Distrib Syst, 2014, 25: 1257-1266.
[15]	Rivest R. The MD5 Message-Digest Algorithm [S/OL]. [2015-01-25]. https://www.ietf.org/rfc/rfc1321.txt.
[16]	KVM Forum. QEMU/KVM [R/OL]. [2015-01-25]. http://www.linux-kvm.org/page/Main_Page.
[17]	Intel Corporation. Iometer [R/OL]. [2015-01-25]. http://www.iometer.org/.

[1]	张磊,张知皦,陈渝. 面向多核虚拟机的高效瞬态协同调度算法[J]. 清华大学学报（自然科学版）, 2014, 54(4): 495-501.