All things related to multicast

[Doug Pearson <>]

My knowledge of multicast and IGMP is pretty limited so far (I bought
the book but haven't read it yet :) but I'm thinking that Steve Francis'
answer to Alan's question is correct, right?

In any event, at IU we experienced a variant transmit-side problem while
testing the Littons:

 > The problem relates to the way Ethernet switches perform IP multicast 
 > pruning - the switch has to hear a "host report" (join) in order to know 
 > that it should prune, otherwise it floods all ports. The fix to this 
 > problem would be for a tranmitting unit to send the host report. Host 
 > reports should be sent initially and periodically at about once every 
 > minute.

Litton provided a small application that does IGMP host reports.
I'll send the app under separate email to Alan.  We worked with Rick
Albertson at Litton to solve this problem... he was very helpful.

- Doug Pearson



At 07:40 PM 12/15/99 -0500, Wallace, Steven S wrote:
>The camvision2 folks have developed a small application that does IGMP host
>reports for the multicast group.
>
>Doug Pearson has more details.
>
>Doug?
>
>> -----Original Message-----
>> From: Steve Francis 
>> [mailto:]
>> Sent: Wednesday, December 15, 1999 6:48 PM
>> To: Alan Crosswell
>> Cc: 
>> 
>> Subject: Re: howto do set up multicast in a pim-sm w/o IGMP?
>> 
>> 
>> Not knowing what a Camvision2 is, I'll assume its a multicast source.
>> 
>> If that's true, you dont need IGMP: a multicast source just starts
>> multicasting, with no prior signalling.  That will cause the 
>> local router to
>> build the multicast tree for that source.
>> 
>> Alan Crosswell wrote:
>> 
>> > I have just learned from Litton that the Camvision2 
>> multicast support does
>> > not do IGMP.  Is there an IGMP proxy available that I could 
>> use?  Am I
>> > correct in assuming that if I'm running pim-sm that I need IGMP to
>> > cause the tree to be built? I am also using Catalyst with 
>> cgmp or igmp
>> > snooping....
>> >
>> > I have my Littons for another 7 days before the eval runs out and
>> > would like to try some IP multicast now that I've 
>> "finessed" unicast.
>> >
>> > /a (still somewhat confused by most of this)
>> 
>