Ars Technica on IPv4 exhaustion

Wed Jun 18 17:40:08 UTC 2014

On Jun 18, 2014, at 09:56 , Niels Bakker <niels=nanog at bakker.net> wrote:

> * mail at martingeddes.com (Martin Geddes) [Wed 18 Jun 2014, 18:17 CEST]:
>> It may never become the defacto standard, period. Nearly 20 years to reach
>> 2% penetration is a strong hint that the costs outweigh the benefits.

The 2% number is also not particularly meaningful. Traffic levels as measured by Google are closer to 4%, but even that doesn't tell the whole story.

The total deployment of IPv6 is probably much closer to 15-25% globally. The astonishingly lower traffic figures are a result of the following likely factors:
	1.	They represent the intersection of client AND servers that are IPv6 enabled.
	2.	They are further reduced by happy eyeballs often preferring IPv4 even when IPv6 would work.
	3.	End user and enterprise adoption is lagging, even where IPv6 could be fully deployed in minutes without any harm.

> Never before have we run out of IPv4 address space, so this time may well be different, now that an actual need for change is developing.

Indeed. A time is coming when new content and services will be unable to be deployed on IPv4 due to lack of number resources. Once that starts to occur, IPv6 becomes the only viable alternative. The question at this point is not whether IPv6 will become the de facto standard, but how much pain we will inflict on the general public in that transition process.

If we deploy IPv6 ubiquitously before we reach that point, then the pain of transition can be minimized. If we fail to do so, then the transition will be abrupt, painful, and very disruptive.

Unfortunately, this is a classic recipe for the tragedy of the commons. We must all act in our mutual best interest deploying IPv6, or, we will all suffer together. Sadly, those who deploy IPv6 later will suffer the least at first and what happens in the long run remains to be seen.

Owen