Do we know the maximum ad-hoc socially acceptable stream size that we can
run across ESnet?
-Matt
On Thu, 30 Jul 1998, Hugh LaMaster wrote:
>
> Well, I stand corrected on one point. Although ESnet can handle
> multicast streams as large or larger than 1 Mbps, because it is
> used heavily operationally, ESnet would rather *not* have us do
> ad hoc testing of 1 Mbps streams. They would like us to notify
> them in advance, and, in general, 1 Mbps is considered "antisocial"
> within ESnet. The reason for this, which doesn't surprise me,
> is that although they have the bandwidth, their mrouted-based
> mrouter infrastructure can't handle loads much larger than that,
> and they have lots of operational users.
>
>
> In short, they would rather work with us to enable a PIM-only path
> from ANL to NREN.
>
> ------------------------------------------------------------------
>
> Joe Burrescia <joeb@es.net> writes:
>
> > [...] sourcing an unbounded megabit multicast stream is
> > pretty anti-social. We do have users that make daily production use of
> > multicast (both on and off ESnet) and this would be disruptive to them.
> :
> > We (ESnet) has been pushing PIM out into our production network, but we
> > are not totally there yet. In the ANL path, we have PIM running from
> > AMES to LBL but then dvmrp to llnl, lanl and finally ANL. We
> :
> > We have run internal high bandwidth multicast streams in the past, but
> > we used administrative scope to contain them. You are right, most of
> > our links can handle this traffic without a problem, but the fanout
> > of the mrouters is the limiting factors. Over running these mrouters
> > will adversely effect the folks who are using the service in a
> > production manner.
> >
> > If you want to run some high bandwith rate tests please let us know
> > before hand so we can monitor what's going on (trouble@es.net is a
> > good place to send this type of mail in case I'm out).
>
> ------------------------------------------------------------------
>
>
> Therefore, before turning up the bandwidth from ANL, we need to
> notify Joe Burrescia <joeb@es.net> and <trouble@es.net>.
>
> And, we need to pursue the PIM-only path anyway; ESnet is willing
> to work with us to enable that.
>
>
> Further additions, corrections, and comments welcome!
>
>
> Regards,
> Hugh LaMaster
>
>
>
>
_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/
Matt Chew Spence Network Engineer/Systems Engineer
matt@nren.nasa.gov NASA Research & Education Network
(650) 604-4550 (voice) Ames Research Center Mail Stop 233-21
(650) 604-3080 (fax) Moffett Field, CA 94035-1000
_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/_/