基于哈希树的分布式目录同步方法

Distributed directories synchronization method based on Hash tree

在云存储应用中,用户通常需要在多个终端上对其工作目录副本进行修改,如何在分布式目录副本之间进行高效的数据同步是一个重要问题。设计实现了一个面向私有云存储的分布式目录同步系统HTD2Sync。系统以文件哈希值为依据进行并发同步冲突检测,能够在文件同步过程中过滤大量伪冲突;针对目录副本同步中的2种冲突类型和6种冲突场景,给出了对应的冲突消解方法;引入有序哈希树对用户目录副本的内容和结构进行建模,给出了有序哈希树的创建和更新方法。提出了一种基于有序哈希树的分布式目录副本同步方法,对其流程和核心操作步骤进行了说明,给出了有序哈希树比对算法COMPARE_OHT。实验表明,HTD2Sync系统能够快速感知私有云存储终端的文件变化,在云端文件变化感知方面也具有较高的效率。

In cloud storage application scenarios, users usually need to modify their working directory copies in different terminals, and then it becomes an important issue for efficient data synchronization between distributed replications of the working directory. Optimistic replication and the eventual consistency theory was used as reference~ a distributed working directory synchronization system named HTD2Sync was de- signed and implemented. In HTD2Sync, file hash value was imported as foundation to detect conflicts in concurrent file synchronization, so that most of the pseudo-conflicts were filtered~ two conflict types and six conflict scenarios in directory replication synchronization were analyzed, and the corresponding resolu- tion methods were proposed; Ordered Hash Tree （OHT） was used to represent the character of user working directory replication, and the create and update methods of OHT were introduced~ the distributed directory synchronization method based on OHT was proposed, and meanwhile, its progress and core op- eration steps were elaborated, and algorithm to compare OHT named COMPARE_OHT was also pro- posed. Experiments show that HTD2Sync can detect client file changes quickly, and also has a good per- formance in detecting cloud file changes.