[net-perf] Getting Started?

[Bob Grant]

Sorry, Russ, I mistakenly replied to you and not the whole list
so here's the thread.

All else equal, we'd prefer to use FreeBSD rather than Linux
which implies BWCTL and OWAMP.  BWCTL looks like the easiest
to implement.  OWAMP has the dependency on really accurate time.
We have some GPS time source hardware but have not installed it
due to needing to run coax up a couple of floors for the GPS antenna.

NDT or Web100 sounds pretty nice for the end user but I question
whether or not it provides as accurate data as the other two.
It requires Linux so we have to pick from the least commercial
of the alternatives ;-)

Finally, I wonder if folks think that one would need separate
hardware for each test?  We have some monitoring hardware that
has dual 2.8G Xeons and dual GigEs coming off a PCI 133Mhz 64 bit
bus.  Its probably overkill but the 32bit 33Mhz PCI is strapped
to do the bidirectional Gig in my experience.  Of course, one
might ask if it is necessary to do Gig with iperf. I'd propose
the answer is yes.

Bob Grant

On Mon, Dec 13, 2004 at 05:12:29AM -0800, Russ Hobby wrote:
> Thanks Bob!  Anyone else?
> 
> Russ
> 
> At 08:40 AM 12/9/2004, Bob Grant wrote:
> >Hi Russ,
> >
> >Russ Harvey and I plan on implementing whatever the group decides
> >is the standard Performance Measurement tools here at UCR and
> >will get it hooked up to HPR.  So far we've been busy with other
> >things and haven't had a chance to look closely at the links
> >you provided earlier.
> >
> >Consider us willing participants but probably not leaders
> >in the process ;-)
> >
> >Bob Grant
> >
> >On Tue, Dec 07, 2004 at 09:37:36AM -0800, Russ Hobby wrote:
> >> Hello?  Anybody home?  ;-)
> >
> >--
-- 
Bob Grant                   |  University of California, Riverside
Director of Technology      |  http://citrus.ucr.edu/~bob/
Computing & Communications  |  rand://cb00ba74a68a83478a0f19baf18fbc1e/