hmm, curious argument - why do you think MOSIX will be more efficient
than a "normal" cluster when confronted with a trivial/serial workload?
is there something about a typical pbs/sge/lsf/etc system that you think
can't handle serial jobs?
I probbably expressed my self i a bad way. What i mean is that with a MOSIX
extension you can start a large bunch of "serial" processes on one node and
these will then migrate in a balanced way in the cluster. A MOSIX extension
sure, Mosix has been around a while, so is reasonably well-know,
as is Scyld's approach. what I'm interested in is whether Mosix
functions well in a more-than-toy cluster (say, at least 100p).
I guess I also am uncertain where Mosix's competitive advantage lies.
my experience is that a serial-job workload is so undemanding that
migration is unnecessary - it's mainly for parallel jobs where you
really want migration. (in SHARCnet, there is a strong correlation
for serial jobs to also be relatively short and quite small in memory
use. not _all_ are, of course, but certainly most.)
regards, mark hahn.
_______________________________________________
Beowulf mailing list, Beowulf@beowulf.org
To change your subscription (digest mode or unsubscribe) visit
http://www.beowulf.org/mailman/listinfo/beowulf
