The coming IP address drought, in plain english

The internet is running out of space. Not disk space or the data transfer capacity they call bandwidth – - we’re growing those handsomely. It’s desperately short on IP addresses, the unique numbers that allow computers to identify themselves on the network.

In my weekly Hands On column in the Australian Financial Review, I explained the problem for non-technical business readers.

What’s it all about ?

The issue is easily explained. If Australia had not switched from standard six digit phone numbers to seven digits, and later eight, there wouldn’t have been enough numbers available to service the explosion in landlines since 1970. Between an expanding population, increasing service penetration and the rise of the fax and modem, the hundredfold increase in available numbers facilitated by an extra two digits was essential.

The same issue has struck the internet. The incumbent system for constructing and allocating IP addresses, the net’s equivalent of a phone number, is called IPv4. It was devised in days when only a science fiction writer would have dreamed that ordinary folk would need an IP address. And not just one address, thanks to the proliferation of portable devices that are internet-enabled.

It will happen sooner than you think

Now for the sharp edge. It seems we’ll run out of new addresses within the next thirty months. After that, patchwork fixes will be required to support growth.

Australia’s chief internet scientist Geoff Huston berated the industry at last week’s Australian Network Operators Group (AusNOG) conference for failing to implement IPv6, a complete solution for the coming address drought. By increasing the number of digits in an internet address, IPv6 expands the range by trillions of trillions.

We had time to sidestep the problem

Just a few years ago, the engineers who architect the net had it all figured out. First, they’d explain the problem and demonstrate the exponential rate at which IPv4 addresses were being consumed. Then they’d unveil IPv6 with its mind-boggling possibilities. Every PC, every toaster, every automobile, every radio, every phone, every DVD player ever sold could have its own internet address. Every traffic control signal, every automatic door, every ambulance. Everything could potentially talk to everything else.

Next, the boffins reasoned, industry would implement IPv6 in good time for an orderly handover from IPv4. By dual stacking internet hardware with both systems, manufacturers could provide a smooth transition well before the old system ran out of space. By, say, August 2008 the net would no longer depend on IPv4.

But we’ve hardly lifted a finger

The problem is that we’re in August 2008 and almost no progress towards IPv6 has been made, prompting Huston to sound the alarm. It’s way too late to introduce it before the current well runs dry, says Huston. To have any chance of a solution, there’ll need to be a nearly religious fervour for change, and we’ll have to buy time using band aid technologies like Network Address Translation (NAT).

Expect to see plenty of NATting

If you run a wireless network at home, chances are it uses NAT already. Your wireless router has a real internet address and can talk with the net. The PCs in your home have pretend addresses that the net can’t see or recognise. The router passes traffic back and forth, keeping tabs of which data stream belongs to which PC on the home network. As a result, you need only one real IP address to connect several home PCs to the internet.

The same thing can be done on a large scale, and that’s our main hope of prolonging the life of IPv4. But it’s a compromise. As anyone who has tried to set up a mail or web server behind a NAT router knows, there are special difficulties. An internet reliant on carrier-level NAT is a poor substitute for the visionary possibilities of IPv6.

The law of supply and demand will raise address values

Huston predicts that the dwindling pool of available addresses will create a market for them, with sky high prices over the next five to ten years. With speculators and legitimate users likely to pounce on any available address blocks from now on, he gloomily predicts irresponsible consumption of the remaining scarce resources. He foresees years of extensive NATting and expensive IPv4 addresses before we bite the IPv6 bullet.

How did this happen ?

How did we reach this point ? The cost of converting to IPv6 mainly falls on Internet Service Providers and carriers. While IPv4 still has some room for growth, few of them have been willing to dent margins or increase prices to fund the operation. It’s like offering an electorate a choice between tax cuts and new dams. If they have never experienced water restrictions, extra reservoirs are a hard sell.