Re: [v6ops] I-D Action: draft-ietf-v6ops-nat64-experience-04.txt

[Simon Perreault <[e-mail]>]

Le 2013-11-20 10:18, Simon Perreault a écrit :
> Le 2013-11-20 03:09, Mikael Abrahamsson a écrit :
>> One way would be to have a recommendation to detect NAT64 and then
>> detect NAT44, and in the case of both being present, prefer one of them
>> consistently?
>
> The current state-of-the-art in NAT traversal expressly discourages this
> kind of one-size-fits-all discovery. You need to think about NAT on a
> per-flow basis, not on a per-host basis. The presence or kind of NAT
> that you will discover may depend on IP source or destination, transport
> protocol, port number, load on the NAT device, time of day, phase of the
> moon, etc.

I forgot to finish my thoughts: but it might be good enough if we scope 
what we want to accomplish with discovery tightly enough. For example, 
using the information for prioritization in happy eyeballs might be more 
than fine.

Simon
-- 
DTN made easy, lean, and smart --> http://postellation.viagenie.ca
NAT64/DNS64 open-source        --> http://ecdysis.viagenie.ca
STUN/TURN server               --> http://numb.viagenie.ca