基于运行阶段特征的虚拟机实时迁移技术

doi:10.11959/j.issn.1000-436x.2016021

摘要/Abstract

摘要：

针对预拷贝算法在起始、迭代、结尾 3个阶段所表现出的不同特点，提出了基于运行阶段特征的虚拟机实时迁移技术(LMCOS, live migration based on the characteristics of the operation stages)。起始阶段引入比对初始内存页的变量传输技术以避免未改变内存部分的传输；迭代阶段引入计数排序传输方法以减少内存页的重传；结尾阶段引入调减虚拟机CPU时间片的策略以缩短停机时间。与预拷贝算法相比，LMCOS使停机时间平均减少53%，总迁移时间平均减少65%。

关键词: 实时迁移, 预拷贝, 运行阶段特征, 虚拟机

Abstract:

Being directed against the different characteristics of start, iterative and end stages of pre-copy algorithm, live migration based on the characteristics of operation stages referred to as LMCOS was proposed. In the start stage, the technique of comparing initial memory page and sending variables was brought to avoid the transferring of unmodified memory. During the iterative stage, the transmitting method of counting sort was brought to reduce retransmitting mem-ory pages. In the end stage, the police of reducing time slices of virtual machine's CPU was brought to shorten the down-time. Compared with pre-copy algorithm, LMCOS reduces downtime by 53% and total migration time by 65% on average.

Key words: live migration, pre-copy, characteristics of the operation stages, virtual machine

邹庆欣,郝志宇,云晓春. 基于运行阶段特征的虚拟机实时迁移技术[J]. 通信学报, 2016, 37(1): 170-179.

xin ZOUQing,yu HAOZhi,chun YUNXiao. Live migration based on the characteristics of operation stages for virtual machine[J]. Journal on Communications, 2016, 37(1): 170-179.

图/表 17

图1

表1

图2

图3

图4

图5

表2

图6

图7

图8

图9

图10

图11

图12

图13

图14

图15

参考文献 15

[1]	CLARK C , FRASER K , HAND S . Live migration of virtual machines. Networked[C]// Systems Design ＆Implementation(NSDI). c2005:273-286.
[2]	HINES M R , DESHPANDE U , GOPALAN K . Post-copy live migra-tion of virtual machines[C]// ACM SIGOPS Operating Systems Review. c2009:14-26.
[3]	陈阳, 怀进鹏, 胡春明 . 基于内存混合复制方式的虚拟机在线迁移机制[J]. 计算机学报 2011,34(12): 2278-2291. CHEN Y , HUAI J P , HU C M . Live migration of virtual machines based on hybrid memory copy approach[J]. Chinese Journal of Com-puters, 2011,34(12): 2278-2291.
[4]	LIU H K , JIN H , LIAO X F . Live migration of virtual machine based on full system trace and replay[C]// Symposium on High Performance Distributed Computing(HPDC). c2009:101-110.
[5]	CHANCHIO K , THAENKAEW P . Time-bound, thread-based live migration of virtual machines[C]// Cluster, Cloud and Grid Compu-ting(CCGrid). c2014:364-373.
[6]	JIN H , DENG L , WU S . Live Virtual machine migration with adaptive memory compression[C]// Cluster Computing and Workshops(CLUS-TER). c2009:1-10.
[7]	ZHANG X , HUO Z G , MA J . Exploiting data deduplication accele-rate live virtual machine migration[C]// Cluster Computing(CLUS-TER). c2009:88-96.
[8]	SVARD P , HUDZIA B , TORDSSON J . Evaluation of delta compres-sion techniques for efficient live migration of large irtual ma-chines[C]// Virtual execution environments(VEE). c2011:111-120.
[9]	MA Y Q , WANG H B , DONG J K . ME2:efficient live migration of virtual machine with memory exploration and encoding[C]// Cluster Computing(CLUSTER). c2012:610-613.
[10]	DU Y Y , YU H L , SHI G Y . Microwiper: efficient memory propagation in live migration of virtual machines[C]// Parallel Processing(ICPP). c2010:141-149.
[11]	SV?RD P , TORDSSON J , HUDZIA B . High performance live migra-tion through dynamic page transfer reordering and compres-sion[C]// Cloud Computing Technology and Science(CloudCom). c2011:542-548.
[12]	ZOU Q X , HAO Z Y , CUI X . Counting sort for the live mi ration of virtual machine[C]// Cluster Computing(CloudCom). c2013:1-5.
[13]	LIU Z B , QU W Y , LIU W J . Xen live migration with slowdown scheduling algorithm[C]// Parallel and Distributed Computing, Appli-cations and Technologies(PDCAT). c2010:215-221.
[14]	JIN H , GAO W , WU S . Optimizing the live migration of virtual ma-chine by CPU scheduling[J]. Journal of Network and Computer Ap-plications, 2011,34:1088-1096.
[15]	ADEL A , KAMRAN Z . Improving the time of live migration virtual machine by optimized algorithm scheduler credit[C]// Computer and Knowledge Engineering(ICCKE). c2014:346-351.

前一轮中的状态	临近后轮的状态	发送
1	1	否
1	0	是
0	0	否
0	1	否

负载	CPU	网速	磁盘	内存
Nbench	90%~100%	0	0	0~5%
Sysbench	0~60%	0	0	70%~80%
Webbench	60%~85%	60~80 Mbit/s	0	0~1%
Dbench	0~5%	0	20~30 Mbit/s	0~3%