All things related to multicast

[David Meyer <>]

        Mark,

        BTW, I monitored the Abilene backbone (and many others) 
        during the event and I can report that multicast on Abilene
        was rock solid. I can say a few additional things about what 
        we learned:

       (i).     Netaid multicast went really, really well.

       (ii).    0% loss inside the UO campus:

cisco7-gw#sh ip mpacket 128.223.156.23 224.2.249.91 quality 
Calculating RTP data quality for 224.2.249.91
  Session: NetAid FrontStage (MPEG1)
  Source: 128.223.156.23 (uovid-00.uoregon.edu), Port: 52972
    Packets received: 302, lost: 0, loss percentage: 0%
    Packets misordered: 0, average loss gap: 0
cisco7-gw#


                This was verified in the client at various sites. 
                BTW, routers in path at UO were running 12.0(6.5).S1,
                as were the OGIG routers.   

        (iii).  Successful sites had *no* additional
                configuration or other problems. i.e., multicast
                just works in the standard model. cf. Berkely, IU,
                NASA, and many others. 30 frames/sec, little or 
                no loss.

        (iv).   Sites with problems were excusively related to 
                ATM (possibly in conjuction with down-rev code). 
                
                o p2mp at MCNC caused auto-rp problem. Switching
                  p2p solved that.

                o LANE at GATech

                  I really didn't get time to fully understand what 
                  was happening at GATECH.

        (v).    L2 Switching

                o HP 4000s at UO have broken IGMP snooping, which
                  had to be turned off (maybe expect router to
                  join the group).
                  
                o CGMP configuration (NASA)

        Thanks again for your input. I'll also plug the UO guys here,
        Hans, Joel, and Lucy. They did an awesome job of getting the
        content up.

        See you in Seattle.

        Dave


According to Mark Fullmer:
> 
> The UO NetAid multicast sessions were a success at OSU.  We had 318
> IP/TV and 27 MacTV downloads, although it looks like there were
> only 106 IP's that connected to the UO IP/TV Content manager on the 9th,
> it still accounts for 27 buildings which were individual routed networks.
> Most of the receivers were students living in the dorms.
> 
> Considering the concert was on a Saturday when most of the faculty
> and staff were not on campus, there was a home football game, advertising
> was only done the day before, and the event was carried on VH1, the
> number of receivers was reasonably high.
> 
> With a 106 active receivers @ 1.5Mb/s the 1.5Mb/s (or two) multicast
> stream(s) vs. potentially 159Mb/s for replicated unicast made this an event
> that would not have been possible without multicast.
> 
> --
> mark
> 
>