GlusterFS
From Wikipedia, the free encyclopedia
Developed by | Gluster Core Team |
---|---|
Latest release | 2.0.0rc7 / 25th March 2009 |
Operating system | Any POSIX-compliant |
Type | Distributed file system |
License | GNU General Public License v3 |
Website | http://www.gluster.org |
GlusterFS is a free software parallel distributed file system, capable of scaling to several petabytes.
GlusterFS is a network/cluster filesystem. GlusterFS package comes with two components, a server and a client. The storage server (or each in a cluster) runs glusterfsd and the clients use mount command or glusterfs client to mount the exported filesystem. Storage can be kept scaling beyond petabytes as demand increases.
To mount GlusterFS file systems, the client computers need FUSE support in the kernel. Servers can run on any node, without any extra prerequisites. As of date, GlusterFS server is tested on Linux, FreeBSD and Opensolaris, and client runs on only Linux machines. 1.3.9 release onwards GlusterFS runs both client and server on Mac OS X Leopard.
GlusterFS is one of the few projects which supports different types of communication transports like TCP/IP, InfiniBand VAPI/Verbs, Sockets Direct Protocol and Unix domain sockets.
[edit] Features
- No Meta-data: Unlike other cluster file systems which addressed the parallelization at block level, GlusterFS engineers believed that problem is at volume management and I/O scheduling level. This enabled meta-data info to be offloaded to underlying mature disk file systems. Backend file system can be accessed directly in case of GlusterFS failures.
- Very modular design, has each features as translators borrowed from GNU Hurd operating system. which can be plugged in depending on the users' requirement.
- Automatic File Replication.
- Can use Stripe translator for getting more i/o performance for large files.
- No kernel patches required, therefore the software can be installed without any server downtime.