Re: Last Call: <draft-weil-shared-transition-space-request-14.txt> (IANA Reserved IPv4 Prefix for Shared Address Space) to BCP

[John C Klensin <[e-mail]>]

--On Friday, February 10, 2012 11:22 -0700 Chris Grundemann
<[e-mail]> wrote:

> On Fri, Feb 10, 2012 at 11:15, John C Klensin
> <[e-mail]> wrote:
> 
>> To follow up on an earlier comment, the rate at which ARIN (or
>> other RIRs) are running out of /10s (or /8s) is probably
>> irrelevant, as are hypotheses about what ARIN staff might do
>> about requests for allocation for CGN use with or without this
>> policy/ block.
>...

> This is not about IPv4 life-support. This is about providing
> the best answer to a difficult problem. Run-out date is not
> nearly as important as efficient use at this point. It is not
> efficient for multiple ISPs to use different space when a
> shared space will function optimally.

Indeed, although I note that it isn't a huge step from the above
to get to "we could really make _much_ more efficient use of
IPv4 space by partitioning the Internet into regions and using
explicit routing or X.75-like gateways to route traffic between
regions, thereby permitting each region to use almost all of the
IPv4 space".  One could even use CGNs to establish that model
with each ISP using almost the entire IPv4 space inside its
castle walls.    Now, I think either of those would be terrible
ideas but, as you and others are probably aware, people who
believe themselves to be serious and competent have proposed
them, almost exactly  

Anything that moves us closer to those models gives me visions
of slippery steep cliffs and the creeps, so I feel a need to be
careful about efficiency arguments too.   YMMD, either because
you see the efficiency arguments differently, because you have
reason to be less afraid of an ultimate "islands connected by
gateways" outcome, or for other reasons.

>> If that difference is less than years, I, personally, don't
>> think that particular argument is useful.   Other arguments
>> may be, but not that one.
> 
> Please see
> https://tools.ietf.org/html/draft-bdgks-arin-shared-transition
> -space for a more thorough analysis of the motivations, pros,
> cons, and alternatives for this shared CGN space. I think
> you'll find that those other arguments are laid out there.

I have and I agree those arguments are there (their interaction
with my fear of the scenarios above is another matter).  Again,
my main suggestion was that we just stop the discussions based
on allocation strategies or IPv4 exhaustion alone... because
they are distractions rather than helpful.

     john