Addressing the protocol problem

The current Internet Protocol can no longer handle the number of computers in the world. Robin Bloor looks at how IPv6 will change all this.

If you have not already become aware of IP version 6 (IPv6), believe me you will. IPv6 will become a cause célèbre similar to the Year 2000 (Y2K) problem.

The major difference is that Y2K was an immovable deadline, whereas IPv6 is a creeping upgrade which will sooner or later create a stampede of implementation.

So what is it? For those of you who do not roll up your sleeves and probe the intricacies of network routing, IP stands for Internet Protocol.

An IP address consists of four numbers each smaller than 256 and separated by full stops; something like 194.169.11.242. In practice, this is a 32bit number.

These four numbers were intended to uniquely identify every computer in the world when the protocol was created, and to act as an address so that messages could be efficiently routed between computers.

Unfortunately, because of the way these numbers were allocated and the dramatic growth of the internet, 32bit was not enough. Indeed, pockets of IPv4 address exhaustion have now started to appear.

Future proofing
So IPv4 is about to be replaced by IPv6, which is 128bit and should last us at least until we start networking our computers with other planets. This address space is large enough to assign a unique address to every proton on planet earth.

You could say that IPv4 was the PC internet, an internet which connected many millions of PCs to many millions of servers. IPv4 was also used within organisations for internal networking, such as networks being isolated for addressing purposes from the internet.

IPv6 is, in theory, the long-term answer to the addressing problem, and possibly a number of other problems. It is not a quick fix for an addressing problem, but a genuine evolutionary step.

Consider security. You do not need to do much investigation to conclude that the big bad network out there is awash with hackers and crackers and other assorted bad guys, all of whom seem intent on engaging in nefarious activity with your network.

IPv6 provides two security enhancements which will act as obstacles for them. Under IPv6, an Authentication Header prevents unauthorised hosts from sending traffic to some destinations by obliging the sender to securely log in to the receiver. This enhancement allows implementers to define the authentication algorithm.

IPv6 also employs an Encapsulating Security Header, enabling the encryption of traffic between two hosts. This is also algorithm-independent. Security is regarded as being so important that you might not have to wait for IPv6 to get it; the IPv6 specifications allow the same features to be added to IPv4.

Mobile IP
IPv6 also provides for mobile IP, which means the ability to assign a unique IP address to every mobile device. In fact it does this by providing auto-configuration capabilities, which is good news for anyone who spends time configuring networks.

IPv6 offers 'stateful' auto-configuration, allowing servers to dynamically assign unique addresses to computers on request using a range of reserved values.

For mobile it offers 'stateless' auto-configuration, where servers generate globally unique addresses by joining the local link address with an additional number.

IPv6 will also be more aware of the traffic that it carries. It will be able to distinguish between, say, real-time video feeds and regular email, and then select the right quality of line and error correction procedures for the traffic.

Very little of the technical story is going to be of much interest to the typical IT user, but the consequences of it all will.

If you imagine a world of hundreds of millions of connected wireless devices (such as laptops, PDAs and MP3 players) and billions of connected things (such as cars, fridges and security cameras) and then imagine all the applications that might run between any of these devices, involving or possibly without the involvement of human beings, and the millions of servers out there, then you have the landscape if not the detail.

Many of the applications of the future will simply not work over IPv4 and, in time, the software producers are not going to even attempt to accommodate it.

For this reason, IT departments cannot wait for IPv6 to happen to them; they need to plan for the future before it arrives. And, as with the Y2K problem, this means all IT departments.

Also, just like Y2K, there will be a cost and there will be no immediate benefit when dealing with the problem, beyond the fact that you'll be able to continue to operate.

And, I suspect, the consultant will make a killing.