面向分布式文件系统的可扩展数据快照技术

A Scalable Data Snapshot for Distributed File System

快照作为保证存储系统数据安全性的重要技术已经得到越来越广泛的应用。然而,在面向海量数据的分布式文件系统中,快照技术在快照创建和快照恢复两个方面存在着高开销和低可扩展性的问题。因此,本文提出了一种面向分布式文件系统的可扩展数据快照技术,采用基于数据压缩的增量快照和恢复链路缩减的方法,实现了可扩展性的快照创建技术和快照恢复技术。通过与目前现有的快照技术相比较,本文从理论角度证明了该方法可以有效降低快照开销,从而实现可扩展的分布式文件系统数据快照。

As a significant technology to ensure the data security of storage system, snapshot has been increasingly used. However, in the distributed file system for big data, it imposes high overload and low scalability both in creating snapshot and restoring snapshot. Therefore, we propose a method of scalable snapshot technology for distributed file system, which implement snapshot creating through incremental snapshot based on data compression and snapshot restoring through reducing the length of link road. By comparing with the existing snapshot technology, we theoretically demonstrated that our method is effective in reducing the snapshot overhead, and implemented a scalable data snapshot for distributed file system.