Without internet addresses, we won’t have an “internet of things”

The people who run the internet are expected to announce (update: they announced it) that they’ve allocated the last set of internet addresses today. In a matter of months, the old addresses based on the Internet Protocol version 4, or IPv4, are going to run out.

That presents a rather challenging obstacle for the visionaries of Silicon Valley who want to create an “internet of things,” where every device has intelligence and is connected to the internet. In this vision, a world full of smart sensors can measure everything and report that data to the internet cloud, which can then make life easier (or scarier) by predicting traffic, heading off man-made disasters, and keeping tabs on everything we do.

If we have no addresses, we can’t activate new devices connected to the internet. Or some of the new devices will not be able to communicate properly with the old devices or web sites. Making Skype phone calls and other means of communication via the web will also get harder. Those are huge engines of commerce that could get delayed, derailed, or stymied because nobody paid attention to the plumbing of the internet. Yes, you could say that we failed to flush our toilets, in a kind of tragedy of the commons where everybody looked out for their own good, to the detriment of all.

The problem is that every new web-connected device has to have an internet address, or the identification numbers, or four sets of four digits, ranging from 0 to 255. When IPv4 was created in 1977, the internet’s creators didn’t think they would need more than 4.3 billion addresses, the theoretical limit that is possible with a 32-bit addressing system. The last five blocks of 16 million numbers each is now being released by the Internet Assigned Numbers Authority, which will dispatch the numbers to five regional authorities that will likely then run out of numbers in six months to nine months, according to expert Owen DeLong at internet service provider Hurricane Electric.

For a long time, the experts who saw this problem coming told big companies that they should switch over to IPv6, which was established as a standard in 1998. IPv6 has an address space in the trillions. But there wasn’t much incentive for internet service providers, enterprises, backbone companies and others to make the shift. There were switching costs involved with no obvious return.

The consequence of running out of numbers depends on how much work those companies now do to head off compatibility problems. We may see disruptions such as the failure of Skype calls to connect because users are on incompatible protocols. That’s because users will have to negotiate communications through a third-party rather than directly connect over the Internet. That introduces complexity and potential slowdowns to the network.

As for the internet of things, we may be a ways off from realizing that anyway. We already have smart, web-connected watches, alarm clocks, cow-tracking system, parking meters, and other sorts of gadgets. But if we want to get to billions of connected devices, we will have to adopt IPv6 en masse and phase out the use of IPv4 networks. For the Luddites among us who think the internet of things is just going to pollute the world with digital nonsense, the “address exhaustion” is a blessing.

The industry has rallied around big causes before, like when it rewrote a lot of software code to get around the Y2K problem, in which legacy computers couldn’t tell the difference between the years 1900 and 2000. Internet giants Google, Facebook, and Yahoo plan to turn on IPv6 for testing on their platforms on June 8, which has been dubbed World IPv6 Day. We’ll have to get through this transition before we can get back to lofty visions of the internet of things.

[picture credit: Tangledwing]