[Gluster-devel] Has anyone... pure nfs replacement

[Craig Tierney]

Brandon Lamb wrote:
> Has anyone replaced nfs with glusterfs using no afr/unify, just a pure
> nfs replacement?
> 
> I ask this as a way to take a baby step into using glusterfs, first
> going from nfs to glusterfs, then when these quirks that are going
> around in recent threads about afr are worked out switching to
> replication.
> 
> BUT, original most important question, anyone seen that glusterfs is
> faster, better than using an nfs server?
> 
> 

I have seen good streaming performance from glusterfs that would exceed
my NFS servers.  However, since I did not run glusterfs on the exact
same hardware, I cannot say for certain that it is better.

I will need someone else to explain this further, but my important
tasks that run on NFS (compiling), NFS is still faster.  For instance,
compiling a current linux kernel takes about 400 seconds on NFS, but
about 1100 on GlusterFS.  I think this has to do with client-side
write caching, but I am not positive.  I have tried different translators
on the client to try and improve the compiling performance, but
have had no success.

I would be interesting in understanding better what is going
on and if it is fixable.

Craig





> _______________________________________________
> Gluster-devel mailing list
> Gluster-devel at nongnu.org
> http://lists.nongnu.org/mailman/listinfo/gluster-devel
> 


-- 
Craig Tierney (craig.tierney at noaa.gov)