A New Data Replication Scheme for PVFS2.

Nianyuan Bao,Jie Tang,Xiaoyu Zhang,Gangshan Wu
DOI: https://doi.org/10.1007/978-3-319-27137-8_35
2015-01-01
Abstract:PVFS is one of the most popular distributed file systems with parallelism, which is still widely used today. Now PVFS is in its version 2, called PVFS2. PVFS2 has a leading performance on I/O operations, but the reliability and stability are not as good. One of the reasons is the lack of data replication. This paper presents a new data replication scheme in PVFS2. In our approach, the backup operation is done on the servers, therefore the user experience is not affected while creating copies of files. In addition, we optimized the read operation of PVFS2. With copies, we can choose the servers to read from, so we can maintain parallelism of read operation under complex conditions such as a server is down or the load of some servers are obviously higher than others. Experimental results verify the effectiveness and efficiency of our method.
What problem does this paper attempt to address?