Re: [Anima] Multicast in the ACP?

[Brian E Carpenter <[e-mail]>]

On 23/04/2015 09:26, Michael Richardson wrote:
> 
> Brian E Carpenter <[e-mail]> wrote:
>     >>> The discovery component of GDNP includes link-local multicast
>     >>> messages.  As far as we know, there is no way to secure these
>     >>> messages, but by definition they are only intended for autonomic
>     >>> nodes, which should all be part of the ACP.
>     >>>
>     >>> So, is there any way to support secure link-local multicast within
>     >>> the ACP?
>     >>
>     >> The ACP as described in draft-behringer-anima-autonomic-control-plane
>     >> is based on a hop by hop tunnel infrastructure. So far those tunnels
>     >> are all point to point. I guess you can send a link-local multicast
>     >> packet through such a tunnel, and it would automatically benefit from
>     >> the security of that tunnel. But of course there is only one other
>     >> node on that "link".
> 
>     > Right, so we would have to specify a "replicast", where the sender
>     > copies the message to each ACP tunnel. That seems like a lot of
>     > overhead.
> 
> http://datatracker.ietf.org/doc/draft-ietf-roll-trickle-mcast/
> describes a way to do multicast across that collection of links... that is if
> we can ever get it through the IESG.

Interesting. That looks like overkill for ACPs that run over non-constrained
networks, but it suggests to me that we need a modular approach, so that it
could be used in networks that call for it.

    Brian