On 2010-10-29 00:24, Keith Moore wrote:
> On Oct 28, 2010, at 3:02 AM, Christian Huitema wrote:
> 
>>> ...  the most workable approach for now requires applications to pass IP 
>>> addresses to peers 
>>> in referrals (perhaps with some additional information) and for those peers 
>>> to make heuristic 
>>> guesses about which addresses to try first.  Granted that's not a very good 
>>> solution, but it's 
>>> way better than trying to prevent apps from making those decisions.
>> Keith, it seems that you are describing ICE.
> 
> not specifically, no.  I'm speaking in general terms here.   given the 
> considerable constraints associated with the current operating environment, 
> different applications will make different tradeoffs.
> 
>> Of course, ICE only really works for UDP based applications. So we may need 
>> to run TCP or a new equivalent protocol on top of UDP. Of course, if we do 
>> that, we can also get a solution that works through not only NAT, but also 
>> basic stateful firewalls. 
>>
>> ICE, STUN, TURN forever. Not exactly what we were shooting for in 1992. Oh 
>> well.
> 
> not what I'm shooting for now either.  but neither do I want to prevent such 
> things from working (for those apps that can use them) in the interim.

I'd invite people who think this topic is of general importance to
read draft-carpenter-referral-ps and send comments to its authors
(or to the [email protected] list). It's proving quite hard to make headway
in getting the IETF to recognise referrals as a problem area in its
own right, but I really think we need to do that, instead of bickering
about the side-effects of various individual mechanisms that break
up the address space or the name space.

     Brian
_______________________________________________
nat66 mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/nat66

Reply via email to