Re: [v6ops] IPv6 multicast and tunnels (RE: v6ops Digest, Vol 52, Issue 41)

["Templin, Fred L" <[e-mail]>]

Hi Bill,

> -----Original Message-----
> From: Bill Owens [mailto:[e-mail]]
> Sent: Monday, December 15, 2014 1:54 PM
> To: Templin, Fred L
> Cc: [e-mail]
> Subject: Re: [v6ops] IPv6 multicast and tunnels (RE: v6ops Digest, Vol 52, Issue 41)
> 
> On Mon, Dec 15, 2014 at 09:10:17PM +0000, Templin, Fred L wrote:
> > Hi Bill,
> >
> > > -----Original Message-----
> > > From: Bill Owens [mailto:[e-mail]]
> > > As of this afternoon, we see 20655 v6 unicast networks and 66 multicast, with paths that include a total of 50 ASes. We have both
> > > commercial and R&E v6 paths, but there are just thirteen peers that carry multicast; all of them are R&E networks. So there is
> *some*
> > > wide-area IPv6 multicast, but not *much*. Certainly not enough to be worth using except in extremely limited circumstances. We
> > > continue to support it simply because of inertia; it would be more work to remove it from our configurations than to keep it in.
> >
> > That is great data, but is a bigger picture concern for IPv6 in general and
> > not for the use of tunnels inside of IPv6 sites right?
> >
> > Inside of the site, there may be a need for IPv6 link- and site-scoped
> > multicasts, e.g., for service discovery, public bulletins, media simulcasts,
> > etc. In that case, it might be good if the tunnels handled multicast.
> 
> I suppose it might be useful to have site-scoped multicast,

Two that come to mind are LLMNR and mDNS. But really, any site-scoped or
link-scoped multicast application should work.

> though if tunnels were needed to accomplish that I think it would be more
> likely that they'd be hand-configured tunnels between routers,

Automatic tunnels without explicit configurations are also possible (see AERO).

> rather than some part of a transition mechanism.

Tunneling is about more than just transition. Also useful for mobility, routing
control, security, traffic engineering, etc. regardless of transition or not.

> I honestly have
> never paid any attention to the prospect of using any of the transition schemes within a site; it always seemed to me that they would
> be between a site and its provider.

Just to be clear, I am considering the entire ISP network as a "site" when
I talk about link- and site-scoped multicast. And the tunnel provides a
"link" that spans the site.

> And although I used to be a strong advocate of multicast both in the wide and local networks, its
> use cases have faded to the point where I no longer recommend that anyone make the effort of deploying it.

Well, there are already RFC4795, RFC6762, RFC6763, etc.

Thanks - Fred
[e-mail]

> Bill.