Thursday, August 12, 2010


The IP is falling! The IP is falling!

Last Friday, one of the talk shows on my local public radio station aired a technology segment about the impending depletion of IP addresses:

Imagine what would happen if the United States ran out of phone or social security numbers. We would be unreachable, even unidentifiable! That is exactly what’s happening with IP addresses: those codes that identify each blackberry, iPad, and Website. IP addresses are set to run out in less than a year. Alexis Madrigal, lead technology writer for joins us to talk about what this means in terms of the way we communicate in an Internet-driven world.

The trouble is that they got a journalist to talk about this, when what they needed was an expert on the technology. Mr Madrigal did a passable job of it, but he missed some key points, exaggerated some issues, glossed over some others, and organized the whole thing haphazardly.

First, we’re not about to “run out of IP addresses.” What will soon happen is that the last blocks of addresses in the current (IPv4, as we call it) system will be allocated. Mr Madrigal does get to that at some point in the conversation, but it’s not stated clearly, and the overall impression given is that well be running out of addresses altogether.

And its not an insignificant distinction. Your Internet service provider (ISP) most likely has plenty of addresses left, and enough to last well beyond the 2011 doom-date. That’s because your ISP has one (or several) of those large blocks, and will continue assigning addresses out of that space for a long time yet.

It’s also because your ISP is using dynamic IP addresses for most of its customer devices. That means that only active customers have addresses assigned, and those addresses are reused when they’re no longer active. Only a portion of a large ISP’s customers are active at any given time.

It’s also because your ISP generally gives you a single address, even if you have several devices in your house or small business. You use a “router” (more accurately, a switch) to connect those multiple devices (two desktop machines and three laptops, let’s say) to your Internet connection (to your cable or DSL modem), using a thing called network address translation (NAT).

Second, there are many (very many) blocks of addresses that are unused or sparsely used, and that could be reclaimed. Mr Madrigal mentions this as well, and it’s a very significant point. Unfortunately, it’s a political nightmare. It will be very difficult to actually do significant reclamation. Even getting back the address blocks for defunct entities will be hard. Grabbing ones back from live companies who see some value in hanging onto them will most likley be impossible.

And so, many Internet standards experts fear an uncontrolled market forming, something that we might look at as a “black market” in IP address blocks, complete with bidding wars and underhanded dealings. Luca Brazzi sleeps with the phishers? Well, something like that.

The shortage of address blocks, along with the broader issue of how many IP addresses exist in the IPv4 system, does put a real limit on the number of devices that will be able to connect to the Internet in the future we view as the “Internet of things”, a time when it won’t just be obvious computer devices, such as desktops and laptops, that will be on the Internet. The set of devices won’t even be limited to iPhones and BlackBerry devices. We’ll have temperature sensors, power meters, thermostats, household appliances, cars, and so on, all on the Internet, each with its own IP address.

At that point, NAT won’t be enough to take care of the problem. At that point, we’ll have to go to IPv6, which opens the IP address space from 32 bits (on the order of 4.3 billion addresses) to 64 bits (on the order of 18.5 billion billion). With another 64 bits for the prefix, we can have 18.5 billion billion blocks of 18.5 billion billion addresses each. That’ll keep us going for a while.

We just have to get the IPv6 deployment expanded. That’s not trivial, but it’s not rocket science either. It’s a question of when it becomes good business to do it. When there’s a real financial incentive to make the switch, the switch will be made.

No comments: