Robert Moskowitz wrote:
> 
> 
> Keith Moore wrote:
>> RJ Atkinson wrote:
>>  
>>> Gentle People,
>>>
>>>   At the 6AI BOF held at IETF last week, there seemed to be
>>> rough consensus on the definitions for these 2 terms:
>>>
>>> IPv6 NAT:    Generic term for any sort of NAT/NAPT/SAT
>>>         for IPv6::IPv6 deployment
>>>
>>> NAT66:        Precise, specific, term for the proposal
>>>         documented in draft-mrw-behave-nat66-*.txt
>>>     
>>
>> I think it's useful to be able to describe not just IPv6-to-IPv6 NAT,
>> but any kind of NAT between IPvx and IPvy.  Since there's already some
>> established use for NAT66 and some other examples of that pattern, I
>> suggest we use the pattern xyNAT for NATs between x and y.
> 
> But aren't 6-6 and 4-6 special gateway cases that have to do special
> things anyway and calling them NATs, while technically correct, masks
> their larger role? They are REALLY needed in the 4/6 transition scheme
> of things.

It may be that there are actually cases where 66NAT is the best way to
solve a valid problem, given the current constraints elsewhere in the
architecture.  (I think the jury is still out on those, but I'm trying
to keep an open mind about them.)

Also, IMHO, if we're going to have NATs in the architecture (I did say
"if"), we need to start encouraging some uniformity of behavior between
them, and work toward providing a predictable, location-independent
programming model for applications.  The current hodgepodge approach to
trying to solve every xyNAT problem separately just threatens to make
the Internet even more unpredictable (i.e. hostile) to applications than
it is now.

So I really do hope to encourage people to think about NATs in general
rather than as lots of different special cases.

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

Reply via email to