The Optimization Of Storage Space In Virtual Machine Backup System

With the rapid development of science and technology, more and more enterprisesemploy the virtual machine to build their own data center in order to meet the growingcomputing needs. However, with the number of virtual machine increasing, theprobability of failure resulting from a single virtual machine failure is increasing.Currently, the primary methods of protecting virtual machines are migration and backup,they are migrating or backing up the complete virtual disk files, which will consume alarge amount of storage resources and network bandwidth. Therefore, to improve thestorage resource utilization is an important issue of protecting a virtual machine.De-duplication is a new kind of data compression technique, which will find identicalinstances of file or data blocks in a storage repository and eliminate all such instances butone. Therefore, the de-duplication based on similarity of data segment not only reducevirtual disk image storage space, but also reduce the backup time sharply. In order to dealwith the problems of indexing and reading performance, HashDB depending on thesimilarity of data segment and VmStore based on the principle of adjacent data storeadjacent are respectively employed.To test the effectiveness of data de-duplication, I conducted extensive evaluations ondifferent sets of virtual machine disk images with different chunking strategies. Theexperiments found that variable-length chunks work well among different virtual diskfiles which come from different virtual platform, while fixed-length chunks work wellamong different virtual disk files coming from the same virtual platform. What’s more,the data de-duplication solution based on data segment similarity can eliminate the samedata effectively, and it can greatly improve the performance of fingerprint matching andreading problems by employing the HashDB and VmStore.