Re: New draft on embedding the RP address in IPv6 multicast address

[ken lindahl <lindahl@uclink.berkeley.edu>]

[ post by non-subscriber.  with the massive amount of spam, it is easy to
  miss and therefore delete mis-posts.  so fix subscription addresses! ]

hi,

At 08:55 AM 10/17/2002 +0900, itojun@iijlab.net wrote:
>>IGMPv2 snooping is enabled now on large parts of campus. i don't think
>>IGMPv3 will play nicely with that. disabling IGMP snooping so that SSM
>>will work doesn't seem like a step forward. where's the hypocracy in
>>that?
>
>        doesn't host compatibility mode behavior (draft-ietf-idmr-igmp-v3-11.txt
>        7.2) help this situation?  caveat: i'm no multicast expert.

and i'm no IGMP expert. but i think the answer is no: it won't help.
if the host(s) drop into v2 or v1 compatibility mode, they won't be
able to send (S,G) membership messages on which SSM depends. that
breaks SSM.

the real problem is the IGMPv2-snooping switch. the switch has to intercept
and process membership messages, that's an essential part of snooping.
if the switch is only v1 and v2 aware, it cannot send the appropriate
(S,G) membership message, so once again SSM is broken.

if, as was implicitly suggested earlier, the switch is somehow smart
enough (or dumb enough) to not recognize the v3 packets and simply
forwards them, then snooping won't work. and that's where i entered
this discussion: i consider snooping a requirement for my campus
network.

if someone who is an IGMP and multicast expert knows that i'm wrong,
i'd be glad to hear it.

ken