计算机工程与设计2024,Vol.45Issue(10) :2881-2887.DOI:10.16208/j.issn1000-7024.2024.10.001

面向KVM虚拟机的无代理连续数据保护方法

Agentless continuous data protection method for KVM virtual machines

陈翰霖 宋亚磊 胡晓勤
计算机工程与设计2024,Vol.45Issue(10) :2881-2887.DOI:10.16208/j.issn1000-7024.2024.10.001

面向KVM虚拟机的无代理连续数据保护方法

Agentless continuous data protection method for KVM virtual machines

陈翰霖 1宋亚磊 2胡晓勤1
扫码查看

作者信息

  • 1. 四川大学网络空间安全学院,四川成都 610065
  • 2. 郑州商学院信息与设备管理处,河南巩义 451200
  • 折叠

摘要

为解决云计算领域中KVM虚拟机缺乏成熟的无代理连续数据保护技术方案的问题,提出一种基于QEMU内部I/O过滤框架的KVM虚拟机无代理连续数据保护方法.通过修改QEMU源代码在QEMU模拟的虚拟机I/O堆栈中插入一个自定义的I/O过滤框架,将虚拟机所有的写数据拷贝一份发送到框架的存储队列,在自定义的库文件中调用框架开放的API接口,获取到虚拟机所有的写数据,通过网络传输到远程备份服务器.实验结果表明,所有类型的KVM虚拟机同一时间点备份前和恢复后的磁盘文件MD5哈希计算值相等,验证了该I/O过滤框架能够在不安装任何代理软件的条件下实现KVM虚拟机的连续数据保护.

Abstract

To address the problem of a lack of mature agentless continuous data protection technology solutions for KVM virtual machine in the cloud computing field,a KVM virtual machines agentless continuous data protection method was proposed based on the QEMU internal I/O filtering framework,named QIOF-CDP.The insertion of a custom I/O filtering framework was introduced into the virtual machine I/O stack simulated by QEMU through the modification of the QEMU source code,all I/O write data of the virtual machine were copied to a storage queue of the I/O filtering framework,and the API interface provided by the I/O filtering framework was called in a custom library file to obtain all I/Owrite data of the virtual machine.The data were transmitted over the network to a remote backup server.Experimental results demonstrate that the MD5 hash values of disk files before backup and that after recovery of all types of KVM virtual machines at the same point in time are equal,verifying that continuous data protection for KVM virtual machines can be achieved by the I/O filtering framework without the installation of any agent software.

关键词

云计算/虚拟化/备份/无代理/连续数据保护/堆栈/框架

Key words

cloud computing/virtualization/backup/agentless/continuous data protection/stack/framework

引用本文复制引用

基金项目

国家自然科学基金项目(U1736212)

出版年

2024
计算机工程与设计
中国航天科工集团二院706所

计算机工程与设计

CSTPCD北大核心
影响因子:0.617
ISSN:1000-7024
段落导航相关论文