We're investigating the possibility of moving GCG from our VAX
cluster to our network of Unix(tm) boxes which share a common filesystem
(Transarc's AFS). We're particularly concerned with the performance
tradeoffs of distributed filesystem access over many workstations vs.
a central (faster?) fs with CPU bottlenecks. Is anyone running GCG
over AFS? How does the NFS performance compare to the central fs model?
Are the tradeoffs worth it?
Is anyone looking at parallelizing GCG? It seems an obvious next step
if you've got many workstations with a common view of the file system...
Thanks.
--
Pat Wilson
Systems Manager, Project NORTHSTAR
paw at northstar.dartmouth.edu
