 |
 |
Hi Mark, I've made limited use of GlusterFS and haven't had any major complaints. Not really enough to qualify as experience for your use case, at least not so far. Cheers, Adam On Tue, Nov 5, 2013 at 3:30 PM, Mark Slater <[log in to unmask]> wrote: > Hi All, > > Thanks so much for all the info! It looks like Lustre may be the way > forward, for the foreseeable future at least. I hadn't thought of the limits > on accessing/writing small files but by default I was going to keep our home > dirs mounted as NFS anyway and just let people know the limitations (saying > that I did have it in the back of my mind to copy the home areas across > though!). I'm interested in glusterfs but if noone else is using it I'm > happier going with stuff that there's experience out there for! > > Thanks again!! > > Mark > > > On 05/11/13 14:23, Sam Skipsey wrote: >> >> NFS4.1/pNFS needs support on the clients, which was historically a >> problem as SL5 kernels were too old to have kernel level support. >> SL6 kernels should be better for this. >> >> Sam >> >> On 5 November 2013 14:20, Ben Waugh <[log in to unmask]> wrote: >>> >>> Where does pNFS fit into all this? >>> >>> Ben >>> >>> >>> On 05/11/13 14:11, james Adams wrote: >>>>> >>>>> Except that the POSIX bit of Ceph is considerably less polished than >>>>> the >>>>> rest of it, at the moment. >>>> >>>> Less polished, but POSIX compliant and fully functional (unless you need >>>> multiple active meta-data servers). >>>> >>>>> Things to rule out: AFS. >>>> >>>> Agreed! >>>> >>>> HDFS is nowhere close to being POSIX compliant and is a bit of a pig. >>>> >>>> - James >>>> >>> -- >>> Dr Ben Waugh Tel. +44 (0)20 7679 7223 >>> Computing and IT Manager Internal: 37223 >>> Dept of Physics and Astronomy >>> University College London >>> London WC1E 6BT