Ars Technica on IPv4 exhaustion

Lee Howard Lee at
Fri Jun 20 20:03:07 UTC 2014

On 6/19/14 11:13 PM, "Christopher Morrow" <morrowc.lists at> wrote:

>On Thu, Jun 19, 2014 at 5:24 PM, Lee Howard <Lee at> wrote:
>> On 6/19/14 4:30 PM, "Christopher Morrow" <morrowc.lists at>
>So, I was focusing on the end-user (Consumer) set because given enough
>migration there that should push more application folk in the right

Some content providers have said that they think IPv4 runout is an ISP
As long as users have IPv4, there's no reason for them to move.
What percentage of eyeballs would need to be dual-stack for app folk to
decide to support IPv6?

>I think ipv6 still suffers from the chicken/egg problem:
>  1) users aren't asking so isps aren't selling/doing
>  1b) ISPs still ahve v4 or a solution (they think) to no-more-v4 and
>can keep rolling new customers out

I simply don't think this is the case anymore, at least in the U.S.  IPv6
deployment to users is huge, and will automatically snowball as old CPE
cycles out.  Mid-sized operators will be coming up this year.  Half of
mobile is done.
I don't know of any U.S. ISP or wireless carrier that is planning to use
the address market or CGN as their exhaustion strategy.

>  2) content places have no one they can't reach today because there's
>v4 to everyone that they care about
>  3) both sides still playing chicken.
>oh well, see you on this same conversation in another 18 months time?

I've said this several times, so for the record, here's my prediction:
After ARIN runs out, and it may be 1-3 years after ARIN runs out, ISPs
will incur the rising costs of IPv4 (through CGN or the address market).
Eventually, costs will be so high that they offer IPv6 at a lower price,
either for paid peering or to consumers.  At that point, content providers
will have a financial reason to migrate, and will painfully find that by
the time they can do so, they will have already lost the users.

To be clear, some content providers support IPv6, and some ISPs support
IPv6.  It's everybody else we need to move. And until they do, the
Internet will be more expensive, or fragmented, or both.

Also for the record: My prediction does not reflect any knowledge of any
specific company's plan.


More information about the NANOG mailing list