I can remember shortly after the year 2000 came and went without a hitch a number of people scoffed at the ‘Y2K’ bug, complaining of the pointless stories of doom and woe which proved to be false. The entire thing was a scam they cried, and the vast sums spent on ‘fixing’ the problem were a waste.

The people who said these things were usually technically illiterate and had very little idea whatsoever what they were talking about. The reason there was no catastrophe in the IT world when the clocks ticked over was down to the tens of thousands of people like me, who worked extremely hard investigating, testing, and ultimately fixing the problems that not having enough bits in date fields caused.

When people tell me there was never a problem, I remember Sunday 2nd January 2000 quite clearly. A panic stricken client called me in to their offices – one of their testing contractors had falsified his Y2K test results over the preceding months. The primary system he was testing didn’t work once the clocks changed, and never could have. I can chuckle about it now, but that Sunday was a rather panic filled and fraught time to say the least.

The next looming IT crisis is approaching, but this time things are rather different. Instead of a fixed date which everyone knows, this time a resource is running out and no one knows quite when it will be gone. You may have seen the odd news article or so about the Internet running out of addresses but you can be sure that these stories will increase at a rapid rate over the next couple of years.

The current computer addressing method is called ‘IPv4‘ . In IPv4, 32 bit numbers represent the address (analogous to telephone numbers) of computers and devices connected to the Internet. For example, an address might be “11000000000010011100100000000001″
and this would represent one unique computer or device with a direct connection to the Internet. These addresses are more commonly written for humans in ‘dotted decimal’ notation, e.g. 192.9.200.1.

There are 4.3 billion unique possible addresses available with IPv4. Unfortunately whilst 4.3 billion may have been adequate in the 1980s when IPv4 was standardised, with the incredible take up and expansion of the Internet over the last 20 years we are on course to run out of these addresses in roughly 2011-2012. Running out of addresses will be a big problem. Companies or users who need an address to connect to the Internet will not be able to get a new one. They will all have been given out.

The replacement for IPv4, is called IPv6 (this is what happened to IPv5 for the curious).

IPv6 uses 128 bit addresses, so a single address might be “110000100001100101001000100100010100000010 1010011100100110101001110000101010100111001 1001001001110000100100110011100100100100101″

A number this long gives 340 trillion trillion trillion possible addresses, which should be enough for a little while at least. In context, it’s enough addresses for every human on earth to have a personal network the size of the current Internet. Or for every one of our home appliances to be hooked up.

We’ve had resource exhaustion problems like this before. Phone numbers in London used to start with ‘01′. A phone number would be for example 01-387-5946.
BT realised that the rapid take up of phones in the capital meant they were running out of numbers, and soon would be unable to allocate a new number to someone requesting one. So in 1990 new prefixes were introduced, 071 and 081.

These prefixes were then superseded in 1995 by 0171 and 0181, and even these were again replaced by 020 in April 2000. Each time these were replaced it was relatively easy to adjust – we humans could just alter the numbers we used to call who we wanted. Dialling extra numbers is no problem to us.

But say for the sake of argument, that instead of extra numbers, new phone numbers had letters added as well. Old phone numbers (such as 020-3128-8600) would be incompatible with new phones which had letters such as ‘3AB40-DA6E3-GB2C1′. Old phones would only be able to call old phones. New phones would need new equipment all the way between the handset on the desk, through the various exchanges used to connect and the phone at the other end.

People with the new phone numbers couldn’t be called by those people with old phones. Every piece of telephony equipment would need to be replaced – phones, office telecommunications switches, fax machines, modems, automated alarm systems, mobile phones. Everything.

This is basically what’s going to happen with the changeover from IPv4 to IPv6. Those computers, phones and network devices will all need to be switched over in order to be able to talk to each other. Companies are starting to look at this problem and are beginning to make the change over to IPV6. However, this isn’t helped when major players in the telecoms arena seem to be rather less than forward thinking.

A temporary fix to the problem of IPv4 address depletion is to use Network Address Translation (NAT). Most companies use this to connect their offices – one TCP/IP address is used to connect many individual computers to the net. This is however just a short term fix.

When the last of the IPv4 addresses is allocated, the deployment of new services will become extremely difficult. The Internet won’t suddenly stop. But there will suddenly be a difference between the digital ‘haves’ and ‘have nots’ with the potential for ‘black markets’ in IPv4 address sales.

Solutions to the migration are mired in complication and cost. For example, Windows XP, though technically IPv6 capable, is incapable of doing IPv6 DNS resolution rendering it unsuitable for use. Mac OSX automatically uses IPv6 connectivity if it’s available, but is incapable of using DHCPv6. The complications of converting an entire Enterprise of systems, hardware, bespoke software and documentation would incur a significant cost. Unsurprisingly, the move from IPv4 to IPv6 has been sluggish.

Some commentators think we will only ever manage a NAT’d IPv4 solution, and the deployment of IPv6 simply won’t happen. The European Commission is trying to encourage us to change over to IPv6, while in other parts of the world different solutions are being rolled out . Companies are planning ahead, though what they are planning is unclear.

Here at Altogether we’re thinking hard about this problem, what it means to us and what it will mean to our clients. It’s very clear there won’t be a clean switch-over any time soon, so with some forethought and planning, the various problems that will occur when the address pool does run won’t come as a shock or surprise.

And that should mean that January 2nd 2011, or September 3rd 2012, or whatever the actual date is, won’t be quite as manic as that Sunday back in 2000 was.

Telephone image by macinate on flickr

Comments