Re: [BEHAVE] RFC6147 and RFC7208 interoperability issues

[Klaus Frank <[e-mail]>]

On 2022-02-07 21:38, Christian Huitema wrote:
>
> On 2/7/2022 9:18 AM, David Conrad wrote:
>> As far as I can tell, Klaus has identified a specific issue in which 
>> existing IETF specifications are deficient in order to facilitate 
>> interoperability in a specific protocol when NAT64 is in use.
>
> Yes.
>
> We have an unresolved tension between recommending end-to-end 
> solutions and recommending more processing in DNS64, which means the 
> IETF will probably need to charter some work if we want to publish a 
> consensus document. But the current discussion yielded both ideas and 
> data.
>
> According to Klaus, practical deployments of NAT64 in ISP use WKP, not 
> NSP. That's interesting. I trust Klaus on that one, and it makes 
> sense: ISP want to avoid breakage, and if using WKP is less likely to 
> cause breakage than using NSP, they will use that. That also means 
> that the job of the application writer becomes simpler: it is much 
> easier to have a special case for WKP than to invent a way to learn 
> the NSP. This could lead to an interesting feedback loop, in which 
> more applications working with WKP means even less incentives for ISP 
> to use NSP. This kind of loop is common and powerful. And maybe that's 
> something that we can document.

 From what I've seen ISPs use both (and while I worked at a cloud 
provider we provided dualstack + WKP to allow customers to build IPv6 
only infra).
WKP in access i.e. customer facing (where DNS64 is necessary) and
NSP in backends for routing and peering/transit (where no DNS is needed) 
but also only if WKP is not an option for whatever reason.

>
> -- Christian Huitema
>
>
> _______________________________________________
> Behave mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/behave