Re: Poll: Both ends behind NAT...

On Sat, Nov 22, 2003 at 09:52:59PM -0800, Gregory Forrest wrote:
> >The current NAT-supporting versions of speakfreely work fine if the
> >NATs are cooperative (which is generally the case).
> 
> Soren:
> 
> I have heard several others refer to "cooperative NATs".  Could you explain
> this term?
> 
> Does this have to do with the NAT box opening particular inbound ports after
> realizing it has an established and known applications running on matching
> outbound (towards the WAN) ports?
> 

I haven't heard the term, but it sounds logical that it could refer to
that.  It seems that most modern NAT implementations handle connection
tracking, and as such they watch outgoing UDP traffic, and
return-route matching incoming packets.   It's not just speakfreely
that needs this behaviour, but a whole series of other applications,
and we just play the game the same way that they do.  It's quite a
logical method, and so seems to be becoming a de-facto standard.

Of course, there may be cases where ISPs are deliberately restricting
traffic using a NAT router, and just provide "web" access.  In that
case, it's not a NAT issue, but rather an access case.  They could
equivalently block ports without a NAT.

Soren


                      * * *

To unsubscribe from this mailing list, send E-mail containing
the word "unsubscribe" in the message body (*not* as the
Subject) to [email protected]