Re: IPv6 only host NAT64 requirements?

[Ca By <[e-mail]>]

On Mon, Nov 13, 2017 at 7:45 AM Ole Troan <[e-mail]> wrote:

> Cameron,
>
> >
> > Yes, the first 2 are not objectionable and are already well know,
> documented and deployed as part of various transition solutions.
> >
> > My feeling is the plane is already in flight and operators do not need
> any additional guidance / confusion from the ietf.
> >
> > In short, this is a solved problem and the industry needs more stability
> and less tweaking .... especially in something fundamental like the node
> definition
>
> This isn't guidance to operators, this is guidance to implementors.


Ok, then i would proceed with caution and weigh heavily the feedback from
implementors



> As we found during the hackathon, multiple stacks do lack the first two.
> And as a consequence applications break.
> Now you can argue if we want to put the burden of this on applications or
> on the host stack. If the former the node requirements document might not
> be the right place.
>
> >
> >
> http://www.worldipv6launch.org/apps/ipv6week/measurement/images/graphs/T-MobileUSA.png
>
> I thought this deployment provided dual stack service? 464XLAT, no?


My definition of ipv6-only includes 464xlat

To your point, a large percentage of the tmobile case is 6555bis with
nat64/dns64.

Based on the picture, you can probably guess when 6555bis was implemented
ands its relative impact



> That's a different problem than what we are trying to solve here.
>
> Cheers,
> Ole
>