All things related to multicast

[Bill Owens <>]

On Tue, Dec 18, 2012 at 07:38:13AM +1300, Sam Russell wrote:
> I'm almost in favour of the opposite - as long as everyone supports
> embedded-RP, nobody in an IPv6 world should ever have to worry about
> inter-domain multicast being hard - no MSDP to worry about.

That certainly makes it easier, though there's still the added complications 
of a second BGP table, PIM, and forwarding state being built in your routers. 
In some ways, native multicast behaves more like another transport protocol 
than an extension of IP (4 or 6). 

There are some important differences in behavior between embedded-RP and 
traditional IPv4 ASM, not least of which is the abscence of a global 
addressing space that anyone can access. I would expect embedded-RP users to 
restrict which sources can access their RPs, for safety's sake; the resulting 
service looks a lot more like a private network. At that point the benefits 
of multicast diminish substantially; unless you're depending on the bandwidth 
savings provided by replicating traffic in the network, I'm not sure there's 
a win. 
 
> Also, since link-local multicast is necessary to make IPv6 work, it
> would seem that multicast would have more support if we'd started off
> with IPv6

That was one of the arguments we used to make, a long time ago, but it's been 
a long time since anyone shipped IPv4 devices without multicast support. 
These days, with the widespread adoption of zero-configuration networking, 
nearly every IPv4 box is using link-local multicast all the time anyway.

> At the same time, it's hard to refute the assertion that "multicast is
> a solution in search of a problem"

Indeed. . . a solution that's been looking for a long, long time, and hasn't 
found its problem yet.

Bill.