The Internet Officially Runs Out of Addresses Today, But It's Not Cause for Panic

IPv4 vs. IPv6

Number Resource Association

Today, in a thoroughly nerdy ceremony in Miami, the last five blocks of IPv4 addresses were handed out. That's sparked a lot of concern: The internet as we know it is out of space! The next evolution of Internet Protocol, IPv6, is largely unusable! Panic in the virtual streets! But that's all a bit misleading, at least for now. Here's what's really going on.

Back in 1981, when the internet was little more than an experiment, Internet Protocol Version 4, or IPv4, was created, in which every computer's identifying address is 32 bits long, allowing for a grand total of about 4.3 billion different addresses. These are typically shown as the IP address we're all used to: Four numbers, each ranging from 0 to 255, separated by dots. PopSci.com is 209.208.1.65. The 32-bit length was basically arbitrary--Vint Cerf, who chose that length all those decades ago, said in an interview, "Who the hell knew how much address space we needed?" More than 4.3 billion, it turns out. ICANN, a group set up by the U.S. Department of Commerce, hands out IP addresses in large clumps to regional internet registries, or RIRs, which represent different parts of the world. Those RIRs then lease their addresses to ISPs, both home (like Comcast) and mobile (like Verizon Wireless).

Today, the last 84 million IP addresses were handed out to RIRs. So first of all, we're not out of IP addresses: Those RIRs haven't even begun handing the last addresses to ISPs, and the head of the North American RIR says he doesn't expect to run out for six or nine months. Then the ISPs have to run out. And then there's the curious fact that, according to estimates, only around 14% of all IPv4 addresses are actually being used, though the remainder are unlikely (due to cost, selfishness of owner, or inherent weird engineering) to free up. (You can read more about that process in CNET's excellent explainer.) All in all, we're certainly not "out" of addresses, not for another year or two--but we are certainly getting close. So what's the solution?

The new version, IPv6, has 128-bit addresses, which gives 2128 IPv6 addresses in total--(hopefully) more than we'll ever need. But you can't just start using IPv6, because it and IPv4 are not inter-compatible. If your computer or smartphone only has an IPv6 address, you won't be able to access websites using IPv4, and vice versa. On the server and network side, it generally requires new hardware or new software to make the transition from IPv4 to IPv6, and that cost combined with the fact that nobody uses IPv6 right now has meant that only about 0.25% of the internet is accessible with an IPv6 address.

The solutions to this problem aren't thrilling. Eventually, we'll probably end up with a "dual-stack" configuration, in which the computer can basically use either an IPv4 or IPv6 address as needed. But you don't have to worry about any of that, really. Modern computers (Windows since XP SP1, Mac since OS 10.2) and smartphones support both iterations, and people generally lease modems from their ISP, so the only gadget consumers own that could conceivably be affected is the router--which is not a huge expense, all things considered. (You can check your compatibility here.) For companies, this is going to be a pain in the ass, at best a tricky one and at worst a tricky and expensive one, but for Johnny Websitevisitor, the worst result is some hiccups in the transition and maybe the purchase of a new router--unless the ISPs decide to pass some of their costs onto him.

That's a possibility, but not a likely one. Consumers hate seeing raises in internet costs, and generally the only way an ISP can make that work is by offering some kind of tangibly better experience by way of explanation. That could be speed (as in Verizon FiOS or 4G wireless) or some kind of feature upgrade, but the switch to IPv6 won't offer anything of that sort to consumers. Compared to other projects, like network expansion, it will be relatively affordable. It's also not something that's easy to explain (take it from me), and it's bad business to raise fees for a behind-the-scenes problem that, if successful, will leave a consumer's experience completely unchanged.