Design and implementation of virtual machine multi-mirror based on fine granularity data separation and fusion

Design and implementation of virtual machine multi-mirror based on fine granularity data separation and fusion

Gailing Guo Kai Shinili

To conserve storage space and facilitate system upgrades, separating the system and user data from virtual machine mirroring is a common practice in cloud computing centers. The current mainstream use is to start the virtual machine with system mirroring, and then to mount the user data as disk, but this coarse-grained fusion method results in the data separation to the user, while the user data cannot cover the system data and the system lacks flexibility. To solve these problems, the principle of starting and using the system mirroring and user mirroring merging is analyzed, this paper presents a scheme of mixing multiple mirror files into a run-time file system, and with the help of UnionFS file system Tools, it realizes the complete blending of multiple mirrors at the file directory level in the KVM virtual machine environment. The experimental results show that the designed expected function is realized, and there is no obvious negative effect on the start-up time of the KVM virtual machine and the reading and writing performance of the file system.

Design and implementation of virtual machine multi-mirror based on fine granularity data separation and fusion