True peers (not clients) behind NATs won't work unless they initiate
the connection anyway, so I'm a bit confused by the model you have in
mind.
On Mar 17, 2009, at 10:36 PM, Bruce Lowekamp wrote:
You're right, I'm referring to inbound connections, which are
necessary to support peers behind NATs, as we're required to do. One
could certainly set up an overlay where NATed nodes cannot be peers,
and so I'm happy to say an overlay should be able to specify such a
configuration, but we can't restrict ourselves to that solution.
Bruce
On Tue, Mar 17, 2009 at 8:10 AM, Henning Schulzrinne
<[email protected]> wrote:
On Mar 17, 2009, at 6:25 AM, Bruce Lowekamp wrote:
Salman,
Unfortunately, the conversation is dominated by the need to work in
the real Internet, where the dominant situation is that UDP will
work
and TCP won't.
I'm curious: Is there data indicating the problem? I've heard of
problems
with *inbound* connections via NATs, but the likelihood that
outbound TCP
from behind a NAT doesn't work seems low, given that this would
essentially
kill all consumer applications, from shiny web 2.0 to dusty IMAP.
Henning
_______________________________________________
P2PSIP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/p2psip
