Re: [MMUSIC] non-SIP ICE: useful or not?

[Rémi Denis-Courmont <rem at videolan.org>]

Le vendredi 18 juillet 2008 17:12:55 Adam Fisk, vous avez écrit :
> I definitely think this is useful.  To me, we need to standardize
> NAT/firewall traversal for as many protocols as possible with as few
> rewrites as possible.

Wait... this goes way beyond the scope of an ICE for non-SIP document... 
because it goes way beyond the scope of ICE.

No joking. If there were to be one standard way to do NAT traversal, that 
would be the client-server model. The server has a public IP, or at least a 
static unfirewalled public IP/port mapping. ICE requires a rendez-vous point. 
In particular, that is why I remain heavily unconvinced of the adequation of 
ICE for RTSP NAT traversal.

Now, I guess (I hope) you were refering to peer-to-peer NAT traversal. Even 
there, we need to keep in mind that ICE has advantages but also nasty 
disadvantages over other solutions, such as Teredo (an IETF standard as well) 
or UPnP (and NAT-PMP).

I don't know the requirements for MIP6 or HIP traversals, but I would expect 
Teredo to provide a lot more natural solution. And then, I wonder what to do 
with TCP (including... RTSP), with ICE-TCP dead-born :(

-- 
Rémi Denis-Courmont
_______________________________________________
mmusic mailing list
mmusic at ietf.org
https://www.ietf.org/mailman/listinfo/mmusic