Re: [Fwd: I-D ACTION:draft-jennings-midcom-stun-results-00.txt

Justin Uberti wrote:

> We use Cisco PIIX for our corporate network. For our ISP customers, we 
> supply fully routable addresses, i.e. we do not NAT them.
> 
> I am curious to find out from our stats how many of our customers who 
> use our software but not our ISP are NATed by their own ISP. Right now 
> we have to go through a lot of craziness to deal with the case you 
> mention where two customers are behind the same top-level NAT but 
> separate residential NATs. Does ICE address this particular case?

Yes. It will dynamically detect the condition and the following will occur:

If the top-level NAT supports hairpin, ICE will end up using a STUn 
discovered address outside of the top-level NAT, i.e., it would take 
advantage of the hairpin support.

If the top-level NAT doesnt support hairpin, AND there was a stun server 
between the residential and to-most NAT, and the clients were configured 
to use such a stun server, those addresses would be used. Note that in 
this case, the stun server doesnt need to be a big, publically available 
server; any user of the same ISP that doesnt own a residential nat could 
run a stun server, and that could be used.

If the top level NAT doesnt support hairpin, and there is no other stun 
server besides the one outside of the outermost nat, then the turn 
server outside the outermost nat would be used.

Thanks,
Jonathan R.
-- 
Jonathan D. Rosenberg, Ph.D.                600 Lanidex Plaza
Chief Technology Officer                    Parsippany, NJ 07054-2711
dynamicsoft
[email protected]                     FAX:   (973) 952-5050
http://www.jdrosen.net                      PHONE: (973) 952-5000
http://www.dynamicsoft.com