... and then "we" say "No." "They" lose credibility and political weight. Next time a decision has to be made, "others" will try to seize power, pointing at this event to show the threat to backwards compatibility that they present.

Politics are everywhere. Don't underestimate the energy necessary to shift the direction of an oil tanker.

They should start by deciding when we're really out of IPv4. We exhausted IPv4 in 2011. And 2012. And 2013. And 2014. Every time somebody says "but this time we're really serious."

There are multiple RIRs who exhausted in various times.

RIPE already ran on the "1024 addresses for every new member and that's it" soft-landing policy for the past couple of years.

But ARIN does not have a nice policy like RIPE does, so it will be a bit more entertaining this time.

None of which makes me care. From the beginning, the relevant question has been, when will I be unable to get an IP? And every time the answer is "real soon now". Yet the price of an extra IP on linode (for one example) has not budged from $1/month. You'll understand if this doesn't inspire much concern.

For the record, I make the same complaint every time. Always the same response. "Those were warning signs, but now the end is nigh!"

When you think of the size of the IPv4 space, you realize that $1/mo for an IP is by far the most profitable add-on they sell.

4.3 billion IPs x $12 per year = $51 billion. If the average cost to actually hold an IPv4 address was even $0.10/month, we would not be talking about IPv4 exhaustion.

My point is, as long as you're willing to spend $12/year just for the IP address, there will be a long line of people happy to sell one to you. But $1 per month for a routable address is at least two orders of magnitude more expensive than network architects are aiming for.

Exactly. And as the home speeds reach 1Gbps to the home, there will not be enough CGN boxes that will be able to sustain that load, so one of the two options:

1) more stateless mechanisms will be used (lw4o6, MAP) - which do not give the entire IPv4 address to residential users => so even if the service has an IPv4 address, the clients might have to go through the hurdles to get to it.

Service providers are planning to go IPv6-only, and content providers already note that IPv6 gives a double-digit performance increase compared to IPv4 [1]

2) The residential ISPs will debundle the IPv4 from the cost - given that a lot of the sites average consumer needs (google, facebook, youtube, netflix) are already IPv6-enabled. For this case, you will not be even able to see that something is "wrong". Just the users will click on the link, see it's not reachable, say "huh?" and move on.

[1] https://www.youtube.com/watch?v=EfjdOc41g0s - 1h12 well spent.

There will be no abrupt end. Just the price per IP will slowly grow, and the usefulness of this IP will slowly fade.

There's no panic. The trick with IPv6 deployment is predicting two years in advance the moment will you need IPv6 "tomorrow" - because for a lot of orgs that's how long it takes.

I'll repost a link from the recent NANOG meeting here about the folks seriously discussing their plans of going IPv6-only:

https://www.youtube.com/watch?v=EfjdOc41g0s

By the way, if you're reading this from a smartphone in the US, try this link: http://test-ipv6.com/

I thoroughly enjoyed the moment when one networking pro (who dealt with IPv6 20 years ago and not since) was telling me he won't have IPv6 till he retires, realized he had his on his LTE connection and was using it for quite some time already! :-)

I repeat myself from another post:

One out of five users hitting google.com and facebook.com in the US today does so over IPv6.

> From the beginning, the relevant question has been, when will I be unable to get an IP? And every time the answer is "real soon now".

It's going to be a long time before you can't get an IPv4 address at all. But what is going to happen over time is the cost of an IPv4 address is going to go up, since demand will increase, but supply won't.

Actually, you may be right, in a way. I've seen some government organizations pull a trick to motivate change - they schedule a change for a certain date... a "hard stop" and everyone has to switch. From the outside this looks like insanity, boiling the ocean... panic!!

The trick comes in, they never had any intention for the change to be complete by that date, just to motivate people with the threat of impending required change. Many will scramble or work hard to meet that deadline, but there are always those who are left behind.

Then, to make it work, nearly before the deadline, the deadline is extended for special cases by a reasonable, margin to allow those that don't convert to catch up.

In any case, the goal is to give the illusion that there is a make it or break it deadline for real, and then address the consequences as they come up after the deadline.

Who is 'they' ?

The organization from the linked site -- ARIN -- American Registry for Internet Numbers.

ARIN has no control over what protocols people use on their networks. They're only responsible for the allocation of new publicly-routable IP address ranges for North American customers.

They've been quite clear and vocal about the fact that they're rapidly running out of addresses and will soon be unable to fulfill requests. There was no urgency when ARIN received its last global /8 allocations four years ago, and there doesn't seem to be any urgency now that we're seemingly weeks away from entirely running out. I don't know what it will take to make people pull their heads out of the sand.