The Internet is full; please go home

Final IPv4 addresses to be given away within days

This is going to sound a little smug, given how little I personally need to do about it and how big a potential problem it may be for other people.

But we all got the news this morning that the Internet Assigned Numbers Authority (IANA) reported that it is down to its last five blocks of IP version 4 addresses and that it will run out completely within the next few days.

I will pause while those readers conditioned by faux-disastrous weather reporting to rush out at the first hint of disaster to run out of the house in a panic to stock up on milk, bread and, apparently, IPv4 addresses.

If this is a disaster it's the slowest-moving, most over-accommodated one in history.

The IETF published the spec for IPv6 in 1998, aiming for a scheme with an almost unlimited number of addresses, compared to the merely appalling number in IPv4.

1998. Do you even remember back that far? You didn't know what a Prius was and if someone offered you a tablet you'd look for the clip that was supposed to hold down the pad of paper.

Twelve years. With almost no progress at all on a migration everyone knew we'd eventually have to make.

Here is a guide on how to get started on IPv6, just in case one like it hasn't been on your screen a million times in the last few years.)

In December of 2008 -- 10 years after v6 was published, a Google study more notable for its casually abusive use of language than any revelations, estimated it was being used on only about 0.086 percent of client computers attached to the Internet. Eight-tenths of one percent.

Why was it not taken up more quickly?

"General worry that turning on IPv6 can cause all sorts of brokenness!!!" according to Steinar H. Gunderson, the Google software engineer who wrote the report, but not the exclamation points.

Other "common" concerns holding up adoption:

  • Tunnels that someone forgot
  • Suboptimal routing
  • Home routers doing evil things to AAAA queries

The real reason no one migrated is that they didn't have to.

Converting to anything is more trouble and more expense than just staying with what you have. If it's a technical thing the cost is a lot higher and the risk of breakage is almost infinitely higher.

Without a compelling reason -- like not having a choice -- it makes no sense to switch.

Remember the Y2K problem? Lots of people would rather their apps stored years as four digits rather than two. Until the milage counter ran close to 1999, threatening to confuse software into thinking it had been transported back to 1900, there was no way to justify the cost of rewriting a big chunk of your software just to write 1999 instead of /99.

Until we run out of IPv4 addresses -- any second now -- there wasn't and won't be a good enough reason to switch. No justification, no budget. No budget, no preparation, and damn little planning.

But you still won't catch a break if the migration isn't easy. It's not a new issue, so even without budgets or time, neither the business-unit VPs nor the CIO is going to have much patience with any big complications.

And tell the truth, even without a budget or long-term plan, somewhere you have a thick file of directions and plans and maybe even purchase orders and reconfiguration guides and probably even some internal hosts or address-mapping/conversion tables already running, don't you?

So you're a lot better prepared for real that it looks like on paper, at least judging by the budgets.

I'll be very surprised if it's not very rare for any decent-sized company to have to do a conversion on the fly with its pants on fire.

IT people don't work like that. They fix problems they know they're going to have, and then peel away the problems to reveal the solution that was there under the covers of the old technology.

So those of us without any responsibility for it will sit back and let those of you who do just fix things and let us know afterward. We'll be aware, in a general sort of way, that there's some very labor-intensive magic being done, but we won't appreciate anything but the results.

And if we don't see any difference, meaning you've done your jobs perfectly, we'll assume it was never a big problem in the first place, just like we did with Y2K.

And you'll go back to planning for the next big catastrophe that will never quite happen, for reasons that will remain a mystery to the rest of us unless we find a pile of insanely detailed contingency plans and diagrams and how-tos and emergency procedures in a thick folder or flash drive that's heavy with data and can actually figure out what the network trolls do with themselves all day.

Probably we'll just put our coffee cup on the pile of paper and go talk about the color of the interface with the people building the corporate iPhone apps.

End users are a little shallow, you know?

Kevin Fogarty writes about enterprise IT for ITworld. Follow him on Twitter @KevinFogarty.

ITWorld DealPost: The best in tech deals and discounts.
Shop Tech Products at Amazon