[arin-discuss] IPv6 as justification for IPv4?

[John Curran]

On Apr 16, 2013, at 11:32 AM, John Von Essen <john at quonix.net> wrote:

> Just for thought....
> 
> Lets say in the future (5 years from now), the entire world has switched over to IPv6 and IPv4 is completely dead in the public space.

A reasonable milestone to consider...  I'll note that it is unlikely that 
folks will immediately reprovision existing working IPv4 customers, so the 
earlier milestone of when the vast majority of content is reachable via 
IPv6 is also of interest, since it is when businesses can stop worrying 
about IPv4 (i.e. they can provision new customers using IPv6, either w/o
IPv4 or with access only to central IPv4 gateway services for access for 
any straggling IPv4-only content) 

> Since v6 space is so huge and abundant, the fees by Arin, Apnic, etc.,. should be almost nothing compared to what they are now since the effort to manage and give it out will be minimal. The blocks are so large, that 99% of Orgs would request one block, and never ever need to make another request again. So the number of support tickets by Arin for resource requests would be a fraction of what they are now. Not to mention, there wont be as many small multi-homed ISP's applying since getting IP space from upstreams will no longer be "difficult".

Agreed. There's still a need for the registry, including various forms 
of access such as Whois, RESTul whois, and then related services such 
as  reverse DNS and RPKI, but the amount of development should drop 
down, particularly if the policy base is stable.  With less requests
for changes, our development workload should be a lot shorter than 
today <https://www.arin.net/features/>

> This means in the future that bodies like Arin will get smaller, with less staff, and a much smaller operating budget.

Correct.  Amazingly, the ARIN Board discusses this possibility quite a 
bit, thinking about that long-term milestones and their implications for 
ARIN's structure and costs.  ARIN's core registry costs still include 
servers, backup, and related system administration tasks even at that 
milestone, but as noted in a previous post, this is only about 1/3 of 
our ongoing budget today.  Even if you add in the ARIN governance and 
same level of activity in Internet Governance, you've only got 50% of 
the costs of today. From a practical perspective, it's unlikely that 
changes in policy and system development will ever truly drop to zero, 
but it certainly could be a lot less than today, with corresponding 
savings in operating budget.

I gave an related estimate on the ARIN ppml mailing list a few weeks back 
<http://lists.arin.net/pipermail/arin-ppml/2013-March/026394.html>, that 
it's conceivable that in a steady state that ARIN's costs on a per ISP basis 
(presently about $2800) could be significantly lower (approximately $1500) 
if one presumes IPv6 success leading to very stable policy and system 
requirements.

> Hmmm, maybe this is why IPv4 is still around, and will remain for a very very long time.

Not ARIN's fault... We've done our share, in that ARIN's services have all
been IPv6 reachable for years.  Get the vast majority of content reachable 
via IPv6, and then your described nirvana is indeed within reach.

> Heck, if we can upgrade every computers OS for Y2K, we can switch the world over to IPv6 and kill v4 once and for all.

Having lived through that comparison for a decade, I'll note that Y2K was
an issue whereby you could test your own systems in advance, and could see
the breakage and fix it in preparation for your next test.  Incentives were
well-aligned with the problem and required steps for solution.  With IPv4
depletion, the problem is that ISPs depend on being able to provision new
customers, but the rest of the Internet doesn't even realize there is an
issue. That is a very, very different situation with respect to incentives.

FYI,
/John

John Curran
President and CEO
ARIN