[Gluster-devel] A big thank you for GlusterFS, comments on roadmap

Brent A Nelson brent at phys.ufl.edu
Tue Mar 20 20:33:53 UTC 2007


1.3 is starting to look pretty solid to me, and I'll be strongly urging 
our big cluster folks here (as well as our campus in general) to try it 
out. GlusterFS is an amazingly flexible and simple/easy piece of software, 
and it already deserves a great deal of attention from the community. 
Word needs to spread, and hardly anyone has heard of it.  Frankly, I hope 
it becomes the de facto standard for networked/distributed filesystems in 
general, not just for HPC clusters.

I will be considering a full department deployment of GlusterFS with 
version 1.4 (with its self-healing so we can do fully automated 
fault-tolerance), although even 1.3 is tempting and a major improvement 
over our existing system.  We'll be drooling over the TLS, quotas, 
snapshots, and ACLs in 1.5, and the dynamic reconfiguration in 1.6 will 
give us long-term peace-of-mind (we'll be able to add and remove nodes and 
disks whenever we need to without disrupting our users, every sysadmin's 
dream).

The encryption support in 1.8 may also be rather appealing, although we
care most about encryption over-the-wire and not as much about encryption
on disk; perhaps that will be provided with the TLS support in 1.5.  On a
related note, has anyone suggested GSSAPI support? This would provide
over-the-wire encryption as well as user-level rather than node-level
shares.

Many, many thanks! GlusterFS is already a dream come true (and getting 
more so on an almost daily basis)!

Brent Nelson
Director of Computing
Dept. of Physics
University of Florida





More information about the Gluster-devel mailing list