All things related to multicast

[Stig Venaas <>]

Frank Fulchiero wrote:
> Stig, Thanks for the informative stats.
> Our SAP server is on 136.244.96.106, and we are sending out SAPs for 3 
> programs. In the future, we may send out some for even more.
> Our SAP Announcement interval is 30 seconds, but the average shown is 
> 9.8 secs, as we are sending out 3 in one program group.
>
> I recall University of Canberra, in the "glory days" of multicast, 
> announcing about 36 international channels.
> There is a tradeoff between being net-friendly and giving viewers the 
> convenience of not waiting to long to see what's playing on VLC, and 
> with other SAP listeners.
> Don't some set-top boxes have these?
> How long after you turn on your TV do you want to see the channel listings?

With my set-top box it does take a while to get the listings initially.
I think several minutes. But then it caches that, even when switched
off. So I rarely notice this.
> What is considered good etiquette? We will gladly decrease our announce 
> frequency to meet that criteria.

I don't have a good answer to that.
> The RFC has a formula for computing announce frequency, but I have never 
> tried to compute it. Just using common sense, and 30 seconds seemed a 
> good compromise.
> I believe in some cases the RFC formula gives an announce frequency of 5 
> minutes, which seems too long in this day and age.

At least with the low usage of SAP and multicast in general, I think 30s
is fine today. It would be a problem if there were a huge number of
announcements.
> How would the stats below differentiate between someone sending one 
> SAP/second for one stream, and someone sending 30 SAPS for 30 streams 
> with an announce interval of 30 seconds?

Right, excellent point. Currently these stats makes no differentiation.
It would be good to have a tool that looked into the actual
announcements.

> Glad you are doing this analysis, seems like it's needed.

Thanks. This was just a quick test. Might do something more if it
seems useful,

Stig

> Frank Fulchiero
> Digital Media Specialist
> Connecticut College
>
>
> On [Jul 24], at 9:24 AM, Stig Venaas wrote:
>
> > We've tried to collect some statistics for sources currently sending to
> > the IPv4 SAP group. See attachment if you like.
> >
> > Right now (when we checked), there was one source (193.10.29.5) sending
> > 10 pps to SAP, which is rather high.
> >
> > Next up there are a handful of sources sending announcement 1pps which
> > also is rather high. Many others at about one packet every 10s which
> > still is rather high.
> >
> > Anyway, see attachment for details.
> >
> > Stig
> >
> >
> > Date 2008-07-24             packet        loss          
> > size(byte)                      packet gap(ms)                
> > thrust(bps)                 source
> > time       span  setup    numb   dup late lost  avg  sdv    avg  sdv  
> > min  max      numb   avg   sdv   min   max      avg    1s 
> > 100ms             file
> >
> > 14:01:53  610.0    0.0       63    0    0    0    0    0    796   90  
> > 669  862         62  9.8s  6.9s   0.0 27.9s     658  7.0K 63.6K  
> > 136.244.96.106:1473->224.2.127.254:9875