Dear Mark,

>>>>> Because of what RFC6333 says, suggesting NOW that solutions that 
>>>>> don't need NATs are variants of DS-lite is a sure way to confuse 
>>>>> people.
>>>
>>> Then we're already confused:
>>>
>>> http://tools.ietf.org/html/draft-boucadair-softwire-cgn-bypass-03
>>
>> Could you explain more which confusion you are referring to?
> 
> Nejc suggested that DS-Lite without NATs (which I assumed meant without 
> a CGN NAT, aka AFTR) would be confusing at this stage.
> 
> The title of the referenced document is "Procedure to bypass DS-Lite AFTR"

For the record, it was Remi who suggested that, I just +1-ed it.

I don't think that the referenced draft really causes confusion, since it 
clearly states what is the "nominal behavior" of DS-Lite: using NAT44 in
the AFTR.

That said, I would be happy too if we could just make two big groups of
all these mechanisms, like "stateful" and "stateless" DS-Lite. However, I
think that is easier said than done. What about stateful A+P solutions like
draft-cui-softwire-b4-translated-ds-lite-01? Is this then stateful or 
stateless DS-Lite?
Compared to 4rd or dIVI, this one has customer-state in the core and if we
consider static port allocation in CGN with DS-Lite, we end up with RFC6333
and draft-cui-softwire-b4-translated-ds-lite-01 very similar amount
of state for both mechanisms. Just the location of the NAPT44 is different.

Nejc
_______________________________________________
Softwires mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/softwires

Reply via email to