一种基于遗传算法的虚拟机镜像自适应备份策略  

A Genetic Algorithm Based Adaptive Strategy for Image Backup of Virtual Machines

作　　者：徐继伟[1,2,3] 张文博[1] 王焘[1] 黄涛[1,2] 

机构地区：[1]中国科学院软件研究所,北京100190 [2]计算机科学国家重点实验室,北京100190 [3]中国科学院大学,北京100190

出　　处：《计算机学报》2016年第2期351-363,共13页Chinese Journal of Computers

基　　金：国家科技支撑计划(2015BAH55F02);国家自然科学基金(61402450);北京市自然科学基金(4154088)资助~~

摘　　要：虚拟机镜像备份是保障虚拟化数据中心服务可靠性的关键技术.同时,为了节省存储空间,镜像备份过程往往伴随着数据去冗余操作.然而,镜像备份和数据去冗余都会占用大量系统资源,会对在数据中心托管的应用性能造成严重干扰.如何缩短备份与去冗余时间以降低对应用造成的性能干扰成为数据中心管理的关键问题.当前常用的备份策略可以分为3种：先去冗余再备份、先备份再去冗余和边去冗余边备份.每种策略都有不同的资源需求,适用于不同的应用场景,而合理的组合策略可以有效缩短备份时间.该文提出一种基于遗传算法的虚拟机镜像自适应备份策略.我们首先针对不同的虚拟机镜像备份策略,分别建立资源需求模型,然后根据系统当前资源占用情况自适应的进行策略规划,以最小化备份时间.实验结果表明：该文所提出的模型可以在2%~10%误差范围内预测去冗余备份时间,采用所提出的组合策略可以减少20%左右的备份时间.Virtual machine （VM） images are frequently backed up for service reliability in data- centers. However, the duplicated data of image backups take up a large amount of storage space. Thus, deduplication technologies are often used in backup operations to save storage space by removing duplicated data. Since backup operations with deduplication are resource intensive and time consuming, how to reduce the time of backup operations has become a key issue of datacenter management. Contemporary deduplication backup strategies can be summarized as deduplication after backup strategy, deduplication before backup strategy and deduplication during backup. As the strategies with different resource requirements are suitable for different scenarios, it is reasonable to combine them adaptively. This paper proposed an adaptive strategy for the deduplication backup of virtual machine images. We first profile the resource requirement of the deduplieation backup operations with different strategies, and then use an object-oriented genetic algorithm to make a plan for minimizing the time of backup operations. Experimental results demonstrate that we can accurately estimate the deduplication backup time, and the algorithm saves about twenty percent deduplication backup time.

关 键 词：虚拟机镜像 镜像备份 数据去冗余 遗传算法 组合策略 云计算 

分 类 号：TP302[自动化与计算机技术—计算机系统结构]