All things related to multicast

[Tim Ward <>]

Quoting Guy T Almes 
():
> M'casters,
> 
>  I would also be interested in comments on Bill Nickless's note.
>  SSM does seem to solve the wide-area source-discovery problem and let us 
> avoid using or improving on MSDP.
>  But I hope that Bill's issue (on the need to improve on our current habit 
> of imposing on L2 switch vendors to snoop intelligently) will receive 
> serious attention.
>  So much of what makes the Internet powerful is the minimal constraints it 
> imposes on L1/L2 protocols.  The result is often wonderful 
> cost-effectiveness resulting from a nice variety of cheap fast robust L1/L2 
> technologies with IP layered neatly on top.  The demands of multicast for 
> snooping seem to go counter to that generally strong attribute of Internet 
> engineering.

I will chime in on this topic.  As many of you know NU is rolling out 
20 channels of broadcast television using IP multicast to our student
dormitories this fall.  Since it's a one way application, it seems like
SSM would be ideal.  But the SSM requirement for IGMPv3 means that we
won't be doing it for several years in that fashion.  The switches that
service our 4500 students were replaced 2 years ago (upgraded from hubs
at that time) and experience suggests that we'll need about 3 more years
of life out of them before we can think about replacing them.  3 more years
before IGMP v3 is supported at the edge of the network.

So while SSM may be technically more desirable, in a campus implementation,
it fails the very real economic test, which to me is at the heart of Bill's 
argument.

Tim