All things related to multicast

["Marshall Eubanks" <>]

On Thu, 4 Sep 2003 14:35:53 -0700 (PDT)
 Greg Shepherd 
<>
 wrote:
> 
> On Thu, 4 Sep 2003, Marshall Eubanks wrote:
> 
> > On Thu, 04 Sep 2003 16:03:28 -0400
> >  Jon Zeeff 
> > <>
> >  wrote:
> > >
> > > How can I find out what are the most popular multicast uses on the
> > > Internet? Or perhaps on Abilene (where netflow data is available).
> > >
> > >
> > > For example, I assume that some of the UofOregon traffic is popular, but
> I
> > > don't know for sure.
> > >
> >
> > Ever since ASM RTCP receiver reports withered and died, no one really
> knows.
> > I am not really sure why this happened. Many people think that having
> > each member of a large group send to every other member is not a good
> thing,
> > but I do not think that it was caused by complaints.
> 
> I recall and early 'complaint' was during the NetAid broadcast. The issue
> was not the sending/recieving of the packets, but the potential explosion
> of state in the routers. Every sending 'receiver' would create an S,G
> entry in the network. We were targeting 10K's of S's for just this one G.
> It became clear that any large audience adoption of mcast would create an
> unmanagable amount of network S,G state IF ASM RTCP were the receiver
> reporting mechanism.
> 
> > My experiments indicate that I do not get RR's for more than a tiny
> > fraction (<< 1/10th) of the audience, but what that fraction is
> > is unknown. Most times all I really know is whether or not
> > I have outbound state.
> 
> Marshall, does your app use mcast of ucast for receiver reports?

The MCTPlayer does send back unicast RR's here, but its use has never been 
more
than a small fraction of the total usage of our MP3 channels (about 1/8th
of reported usage, which is of course not the full usage).

Our new video multicasts use Quicktime (and Real, as soon as we get
the bugs out), and those use standard RTCP.

Marshall

> 
> Greg
> 
> > Marshall
> >
> >
>