Design And Implementation Of Clustered NAS System Performance Improvement And Its Traffic Statistics System

With the rapid development of Internet technology, data storage has also entered the era of big data, the traditional Scale-up architecture of storage technologies such as Storage Area Networks(SAN) and Network Attached Storage(NAS) in terms of performance and scalability on a massive data has been unable to meet application needs.For these requirements, this paper focuses on the current clustered NAS technology, chose the open source cluster file system GlusterFS, CIFS server Samba, high availability components CTDB build a distributed clustered NAS storage system. But in the process of build we found: because GlusterFS is implemented based on FUSE file system, access data through clustered NAS via CIFS protocol will cause multiple core switches overhead, this will affect the performance of the system, for which this paper has developed a Samba VFS module that can called GlusterFS libgfapi directly to provide data access, and designed Gluster FS metadata cache layer, thereby reducing system overhead and improve performance clustered NAS system.In the application, the system administrator can also exist on the demand of statistical users usage, the existing clustered NAS system does not have a ready-made solution, we proposed a traffic statistics system based on log analysis for this purpose of this article, and we also developed the corresponding call interface, developer can make flows statistical by calls the correlation interface directly.Finally, comparison of Clustered NAS system performance by performance testing before and after optimizing, verify the validity and the improved amplitude of system performance.