[arin-ppml] Encouraging IPv6 Transition (was: Clarify /29 assignment identification requirement)

[Owen DeLong]

>> 
>>> But, I didn't say it was risk of collision with ULA-R that was the
>>> main problem, it is lack of reverse DNS and lack of whois that is the
>>> problem.   
>> 
>> Why do you need non-local RDNS and/or WHOIS for local-only addresses?
>> 
>> If the addresses should not be seen outside of your organization, why
> would you need a directory service to tell you who the addresses belong
> to? 
> 
> They *can* be seen in SMTP "Recieved From:" headers.  If it's a v4 RFC1918
> address, it could have come from anyware.  If it's a v6 unique PI or PA
> address, even if from a non-routable subnet, you can at least track it
> back to the assignee.  If it's v6 ULA with no RDNS, you can't tell where
> it came from. 
> 

So, at worst, you are in the same boat with ULA as with IPv4 RFC-1918. Clearly
the enterprise world has deemed that mess as an acceptable one.

Personally, I think ULA is a really bad thing overall and that GUA with registration
makes far more sense. If you don't want it outside, filter the routes and the packets
at your borders.

> There may be other examples where internal addresses leak out into the
> wild.
> 

Indeed, but, unless you can show a way in which the IPv6 ULA situation is worse
than the current IPv4 RFC-1918 situation, then, I fail to see how this is in any way
a reason not to deploy IPv6.

The original claim I was responding to was that in order to deploy IPv6, enterprises
need non-public addresses. ULA meets that test at least as well as whatever they
have in IPv4.

Owen