Re: [Magma] Joining to SSM sources that are IPv6 mobile nodes

[Rolland Vida <Rolland.Vida@lip6.fr>]

Dave et al.,

Solution 2 is obviously the simplest one. However, solution 1 might prove
to be not as bad as it looks like. Brian said "Let the
group members think the MN is always at home sourcing the multicast data."
In fact, you could use your proposed solution 1 without the receivers being
aware of the MN's address change. The MN should advertise its new CoA to
the edge routers, and not to the receivers. Edge routers then will send a
PIM-SSM join message to the CoA, and a second SSM tree is built. This will
be done transparently for the receivers.
There's of course the problem that even if data will arrive on the new CoA
SSM tree, receivers will drop the packets as the source address is not OK
for them. A possible solution for that is to do NAT in the edge routers,
and change back the source address to the home address, before forwarding
the data to their receivers. I agree however that this solution is a lot
more complicated than a simple tunneling.

> Either option (1) would never be used for any multicast traffic, or else
> the host would
> have to know whether a receiver might want to use an include-mode list
> with MLDv2 and tunnel to the home agent if so.

You can make the difference between an SSM and an SM connection by the
group's address. There's a specific SSM address range, and it is supposed
that you do not do PIM-SM  in the SSM range.

More generally, I think that the problem has a lot in common with the
problem of reliability in an SSM environment. Imagine that you have an SSM
source, and a potential repair server somewhere downstream. If data gets
lost, only the source can send repair data. If you want to use the repair
server, you should probably create a second SSM tree, rooted at the repair
server. In addition, repair servers could move, or in fact could be
activated or deactivated in certain conditions. There are some papers on
this issue by people from UMASS. The point is that you will have to
constantly join and leave different SSM repair trees, like the different
CoA trees. Somehow similar, isn't it?

Ok, I'm a bit far from the strating point. But it's an interesting issue I
think

Regards,
Rolland Vida.