【摘要】 针对目前在集群文件系统中出现的元数据的故障问题,在PVFS的基础上提出了一种新的元数据检查点的日志管理策略。该策略在Linux环境下实现,解决了文件系统中较慢的元数据管理这一瓶颈问题,并且具备了较强的容错功能。该方法采用磁盘日志和内存日志的结构,通过对事务的管理,能满足集群文件系统中元数据高可用性的要求。更多还原

【Abstract】 Based on the failure problems in the metadata of cluster file system, a novel metadata checkpoint journal system on PVFS was researched and developed. This system was fully implemented in the Linux environment and could solve the bottleneck in metadata manage of file system, so it was provided with fault-tolerance function. This method used the structure of disk journal and memory journal, and adopted a mechanism of transaction, so it could meet the request of high availability in file system metadata.更多还原