It’s only the most significant architectural development in the history of the Internet, and presto, it transpired last night at 00:01 GMT. Did you notice?
I’m betting not, and that you probably didn’t even know it was happening, which is precisely how things were supposed to go down. Don’t worry, you’re fine, you don’t need to do anything, and as far as most of the Internet is concerned, turning on IPv6 — of tectonic caliber at the architectural level, minus the earthquakes — won’t impact how you interact with the Internet any time soon. But it will eventually. And it was necessary, to prevent the Internet from running out of real estate.
Thus “World IPv6 Launch,” which is what participants have dubbed June 6, 2012, the day some of the world’s biggest Internet service providers and companies like AT&T, Cisco, Comcast, Facebook, Google, Microsoft and Time Warner Cable, enable IPv6 permanently on their hardware. It’s the followup to World IPv6 Day, which occurred a year ago on June 8, 2011, when providers turned on IPv6 for a single day in a kind of symbolic “time to pay attention to this” act.
I heard a lot about IPv6 back in the late 1990s, while working as a network engineer at Union Pacific Railroad. “We’re running out of IP addresses!” — the mantra then, as now. It turned out the sky wasn’t quite falling back then, but it may well have, had today’s switch not occurred.
One of the developments that forestalled an earlier shift to IPv6 was something called Network Address Translation, or NAT, which is probably how you get on the Internet at home or work — one device gets a public IP address, then doles out private IP addresses to everyone behind it, allowing everyone to have their own private network identity in the context of their “local area network” while piggybacking, so-to-speak, on that single public IP address. When my family uses computers at home, for instance, we each have separate local/private IP addresses, but we’re sharing the same public IP address as far as the Internet is concerned. It’s not unlike the way physical mailing addresses work: one shared address, city, state and zip, with only the addressee’s name changing. Without NAT, i.e. if every device had to have its own public IP address, we would’ve run out of addresses a long time ago.
But even with NAT, we’ve run through our IP address trove. It happened on February 3, 2011, in fact, when the Internet Assigned Numbers Authority (IANA) exhausted its pool of unallocated IP addresses. And Cisco projects that in four years — by 2016 — the world will leap from 10.3 billion network connections (in 2011) to 18.9 billion connections. The more, the share-ier: We’re already way past our IP address allocation ceiling, which means forced IP address sharing, say one for an entire neighborhood instead of per house, and consequent access slowdown, to say nothing of the potential security issues.
IPv4, which is how we’ve been doing business — a 32-bit address consisting of four octets separated by decimal points, e.g. 188.8.131.52 — only allows a total of 232 addresses (just under 4.3 billion). Tally up all the people in the world with single or shared Internet equipment that needs at least one public IP address, as well as all the enterprise-level (private, government) systems in need of the same (every website, for instance), and you can see where the 4.3 billion ceiling was bound to be a problem. Enter IPv6, which uses 128-bit addresses, allowing up to 2128 addresses (a number with too many zeroes to write out, but about 3.4 x 1038) — the leap in addressing possibilities from IPv4 to IPv6 is literally exponential.
One problem involves connecting a device with an IPv4 address to one with an IPv6 address, because the two protocols don’t talk to each other without translation devices — devices which naturally introduce latency to the process.
I spoke with John Curran this weekend, the president and CEO of the American Registry for Internet Numbers (ARIN), a nonprofit corporation charged with coordinating IPv4 and IPv6 address distribution. By turning up IPv6, Curran says we’re essentially creating a second Internet — one that the current version can only talk to using translation hardware. It means that once this ball really gets rolling and users start to move over to IPv6 native, businesses still running on the older IPv4 network will be dependent on translation hardware when interfacing with IPv6 consumers, which could impact the performance of a customer experience, say the service involves network-intensive processes (e.g. high-definition video streaming). Switching to IPv6 sooner than later is thus an imperative at the enterprise level, if those businesses don’t want to fall behind, argues Curran.
Want to know if your equipment’s been upgraded to IPv6 already? Google has an IPv6 tester you can run by clicking here. As Google’s self-described “Chief Internet Evangelist” Vint Cerf notes, if you want to know when your ISP’s planning to upgrade you to IPv6, you’ll want to contact them directly.