It really depends on what your exact workload is. I would assume your workload will be large sequential reads/writes or streaming? Myself and the group at AmesLab have done similar things on fast hardware and been pretty pleased with the results for streaming performance.
Can you describe your workload a little bit more? How many clients? How many servers? Type of file-access/pattern? (Random/Sequential?) As far as whitepapers, check out the pvfs.org website documentation page for success stories. Good Luck ! On Tue, Oct 7, 2008 at 5:50 PM, Brian Krusic <[EMAIL PROTECTED]> wrote: > Hi, > > Traditionally in CG (computer graphics), we've used some sort of NAS (like a > NetApp), or a more clustered approach (like an Isilon) to achieve the need > I/O in delivering complex jobs (commercials, movies, video games). > > I've always been intrigued by the PVFS project and want to try it in my env > (pure Linux env). > > Trying something like PVFS is a commitment and I would like to know if this > is applicable in my env. > > A summary of our current env; > > - Linux desktops running Maya, Metal Ray, Nuke, Shake (high end CG apps). > - Gig-e to some kind of slow NAS or JBOD server over NFS. > > Does any one have a white paper on how PVFS is actually implemented in there > env so that I can see if its applicable? > > We are thinking of spending $150K on a NetApp, BlueArc or Isilon and I would > like to look at alternatives rather then going status quo. > > Thanks in advance, > - Brian > > > _______________________________________________ > Pvfs2-users mailing list > [email protected] > http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users > -- Kyle Schochenmaier _______________________________________________ Pvfs2-users mailing list [email protected] http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
