[Beowulf] Input Sought: "Basic" Luster FS deployment on GigEther-Fabric Cluster

Robert Latham robl at mcs.anl.gov
Fri Apr 20 08:02:29 PDT 2007


On Wed, Mar 28, 2007 at 03:30:30PM -0400, Tim Chipman wrote:
> I'm looking to soon begin deployment of a ~50node (dual socket, dual
> core opteron) cluster with gig-ether interconnect, intended to run a
> fairly specific CPU intensive MPI model which scales "absurdly
> well".  I/O performance is far and away not the bottleneck. (based
> on benchmarks done on other linux clusters already)

Do you have a handle on what sort of I/O this application is doing?
The choice of parallel file system (or choice not to use one ) depends
a lot on the workload.

You might want to take a look at PVFS (www.pvfs.org), depending on the
kind of I/O your application is doing.  MPI jobs which are I/O bound
is one of PVFS's target audiences.

==rob

-- 
Rob Latham
Mathematics and Computer Science Division    A215 0178 EA2D B059 8CDF
Argonne National Lab, IL USA                 B29D F333 664A 4280 315B



More information about the Beowulf mailing list