[arin-ppml] NAT444 rumors (was Re: Looking for an IPv6 naysayer...)

Owen DeLong owen at delong.com
Fri Feb 18 10:05:45 CST 2011


On Feb 18, 2011, at 7:34 AM, Zed Usser wrote:

> --- On Fri, 2/18/11, Owen DeLong <owen at delong.com> wrote:
> 
>>> Now correct me if I'm wrong, but isn't some kind of
>> NAT/PAT going to be required to join the IPv4 and IPv6
>> domains in all foreseeable futures? If so, aren't we going
>> to have to deal with these issues in any case?
>>> 
>> No, we need to move forward with IPv6 on all levels in
>> order to reduce the need for these solutions.
>  Reduce, yes. Remove, no. Without a global cutoff date for the IPv6 transition, it's not like IPv4 is going to disappear overnight. Furthermore, without any IPv4/IPv6 translation, the first IPv6 only networks are going to be awfully lonely. 
> 
That depends on the number of IPv4 only networks vs. dual stack networks when that happens.

>> Joining the IPv4/IPv6 domains doesn't work out all that
>> well and a dependency on doing so is
>> broken in a number of ways, many of which are documented in
>> the draft.
>  We agree that IPv4/IPv6 domain interoperability is broken, but it's not like we can ignore the issue. So, unless I'm very much mistaken, the NAT/PAT issues are going to have to be dealt with. Or do you propose an alternative solution?
> 
Dual stacking all the IPv4 networks is the alternative solution. Initially it will be the IPv6 only users that are lonely.
Relatively quickly, it will be the IPv4 only networks that are lonely as the bulk of users will, I suspect, become
IPv6 preferred relatively quickly once there is no more IPv4 at the RIR level.

> Please note that this is not an anti-IPv6 stance. To me it looks like the problems plaguing NAT444 need to be solved just to make IPv4 and IPv6 co-exist. Perhaps not the very same problems, but similar NAT/PAT problems in any case. Please do tell me I'm wrong. Bonus points for explaining why I am wrong or how the IPv4/IPv6 thing is to be solved without NAT/PAT.
> 
I think that effort spent trying to solve those problems is better spent moving existing IPv4 things forward to
dual stack. You only need to solve those problems to the extent that there are meaningful things still
trapped in an IPv4-only world. Move them to dual stack and the problem goes away.

Owen





More information about the NANOG mailing list