[arin-ppml] Proposal insanity --- an open letter

[Milton L Mueller]

Thanks, Paul, for weighing in in support of my point. I am not sure I would call heavy reliance on NAT "pretty straightforward", but it does mean that operators can minimize or avoid NAT by acquiring more ipv4 blocks in the near-intermediate time frames. Hence the importance of transfers and other policies, such as the rationing one you mention, in the ipv4 space. So we are agreed, don't tell people we can ignore or stop talking about ipv4 address policies.

> -----Original Message-----
> From: Paul Wilson [mailto:pwilson at apnic.net]
> Sent: Tuesday, February 22, 2011 12:01 AM
> To: Milton L Mueller; arin-ppml at arin.net
> Subject: Re: [arin-ppml] Proposal insanity --- an open letter
> 
> I don't often post to PPML, but here goes.
> 
> --On 21 February 2011 11:24:23 PM -0500 Milton L Mueller
> <mueller at syr.edu>
> wrote:
> >
> ...
> > Also, to say that ipv4 is "baked" and therefore we should stop talking
> > about it or developing policies for it misses the crucial fact that no
> > one can jump to pure ipv6 without cutting themselves off from most of the
> > internet. "Implementing ipv6" in reality means "implementing dual stack"
> > - now and for the next ten years at least. How do you do dual stack
> > without any ipv4, do tell?
> 
> First: In this context, "dual stack" is defined as including IPv4, of
> course. And the IPv4 addresses which are used can be public or private, of
> course.
> 
> Milton, I guess you understand private addressing and NATs.  Dual stack
> during the transition will involve mostly private IPv4 addresses, and NAT
> of various kinds between those private addresses and the public; along with
> public IPv6 addresses of course.  The more that this NATing goes on, as the
> Internet grows, the less efficient that IPv4-based connectivity will
> become, to the point of being less preferred than IPv6.  At the same time,
> IPv6 connectivity steadily improves in terms of services accessible,
> routing efficiency, and general reliability; starting with today's
> relatively meagre coverage, until it is preferred, first in some, then in
> most, and then in all, places.
> 
> As for the public addresses needed by ISPs for their gateway into the IPv4
> Internet, some RIRs already have a policy which will make this available in
> the long term, effectively rationing some proportion of their remaining
> IPv4 address pools.  APNIC has such a policy in place, currently reserving
> most of a /8 for allocations in /22 blocks; and this supply of addresses is
> projected to last for many years, beyond the point where they are actually
> needed.
> 
> That's the model in a nutshell.  Crack away at it by all means, but it is
> pretty straightforward.
> 
> Paul Wilson
> APNIC
> 
> 
> __________________________________________________________
> ______________
> Paul Wilson, Director-General, APNIC                      <dg at apnic.net>
> http://www.apnic.net                                     +61 7 3858 3100
> 
>