A race to feed the web's voracious appetite
The race for web addresses on the information superhighway is causing confusion on dealing with the problem, says Fiona Harvey
Published: December 6 2000 18:50GMT | Last Updated: December 6 2000 18:57GMT
There are about 6bn people in the world. They own an estimated 350m computers, 480m mobile telephones, 600m cars and as many as 1bn television sets.
Sooner rather than later, all these devices and more will want to be on the internet. Phones will be using the web for mobile commerce, which is predicted to be worth $16bn (£11bn) in 2005, while cars will need to be connected to the net to take advantage of services such as global positioning. Meanwhile, technology companies are working towards a future in which domestic goods such as refrigerators, television sets and video-recorders are connected to networks that householders will be able to access remotely over the internet.
The problem is, there is no room. For any object to be recognised on the internet, it must have a unique identifier, an "address". And the web is running out of addresses.
The technology on which the internet is built, known as the internet protocol, allows for only about 4bn addresses (see below for an explanation of how this system works). Four billion seemed like a huge number to the computer scientists who drew up the protocol in 1981. They did not foresee the explosion of wireless technologies that means every electronic device needs an internet identity. And even if they had foreseen it, they might well have assumed that we would have the time - and the will - to overhaul the infrastructure before the addresses ran out.
The technology for that overhaul has been available for six years. Internet protocol version 6, the upgrade to the present version 4 (version 5 never made it), would allow for a theoretically almost infinite number of addresses. Yet it has not been adopted and may not be for some time.
Computer programmers have an unfortunate history of failing to plan for the future. Remember the millennium bug? All but the most pessimistic technologists must have foreseen that the year 2000 would occur and that, when it did, programs relying on a two-digit date would become obsolete. Yet as late as 1998 most companies had done nothing about it.
It is not just the programmers' fault. Cost-cutting company executives often resist future-proofing their computer systems when a cheaper work-around can be found, even if it is only temporary. Given the pace of technological innovation, companies are also understandably reluctant to commit resources to an untried technology that may never take off.
So, as with so many technology projects, the story of IPv6 is one of delay, fudge and bickering. While leading figures such as Vinton Cerf, who helped draft the original internet protocols, have called on companies to implement the new protocol quickly so as to speed up wireless commerce and the new economy, others have argued that we can retain the present system indefinitely.
The plausibility of this idea depends on where you sit. Internet addresses were unfairly allocated by the Internet Assigned Numbers Authority in the US. Organisations that applied early were given a large number of addresses that were often superfluous to their needs. Those that moved on to the internet later found it difficult to get enough. As American companies were first to realise the commercial value of the internet, they were allocated a disproportionate number of addresses. European companies have fewer; countries such as China and Japan will require far more addresses than they have received.
"There is a huge imbalance between the US and the rest of the world on this and while American internet companies may not see the need to move to IPv6, for the rest of the world the need is more pressing," says Stan Schatt, vice-president of research at Giga Information Group.
Future developments in wireless commerce will make the need more pressing still. At present, many internet service providers have managed to expand their allocation of IP addresses artificially, by allocating them on a temporary basis. Thus, instead of a computer being permanently identified by a set address, it receives a new address from the ISP each time it logs on to the internet. Similarly, each time it logs off it returns that address and becomes unidentified. As long as the number of computers logged on at any time does not exceed the ISP's pool of addresses, the ISP can get by on a smaller block of address space.
However, third-generation mobile phone networks will call for phones to be connected all the time, necessitating permanent addresses. Mobile operators will need to implement more complex methods of managing their stock of addresses.
In Europe, where wireless technology is forecast to take off more quickly than in the US, operators will be severely affected. As a result, says Mr Schatt, "much of the impetus for IPv6 is coming from European [telecommunications companies], which have tended to adopt the new protocol".
Scott Kriens, chief executive of Juniper Networks, which supplies the networking equipment on which the internet runs, reports that mobile operators in Japan are also clamouring for the new protocol. "Until one adds the impact from the device count in the mobile market, [sticking to IPv4] is probably a defensible approach. But the 600m mobile users [we believe will be] on the internet by 2003 will mean we have to go to IPv6," he says.
The cost of moving over will be high. Although most new networking equipment should be engineered for IPv6 - indeed, for an internet service provider to install new equipment based on the old protocol now would be a bit like someone commissioning a non-Y2K compliant computer system in 1999 - most of the installed base is still built on the old technology. To install a new router, which is one of the pieces of hardware over which internet traffic passes, could cost about £250,000 ($362,500). An internet service provider, or a large company, may have dozens of routers.
The analysts at Giga believe that by 2005 IPv6 will take over from IPv4 as the dominant internet protocol. By 2009, only pockets of IPv4 will remain.
There is a strongly opposing argument, however. It states that technologists have found enough work-arounds to expand the scope of the IPv4 protocol so that the addresses it allows for will meet all our future internet needs.
Neil Rickard, research director at the Gartner Group of market researchers, says firmly: "We don't think many people will switch. Certainly not for the next five years."
Mr Rickard argues that users can have their device addresses converted to IPv4 addresses as their transmissions pass through their service provider's network to the public internet.
One possibility is of a hybrid system that contains islands of IPv6 in a sea of IPv4. When transmissions are contained within an IPv6-enabled network - say, from one mobile phone to another, running on the same operator's network - they could gain the advantages of the IPv6 protocol, such as the larger number of addresses.
However, when the transmissions have to be routed over the public internet, which would mostly be using IPv4, they would have to be adapted to the old IPv4 address system, using a piece of conversion software as they left the internal network. As long as there are still a number of IPv4 users, companies wanting to have access to the whole internet will be obliged to comply with IPv4.
In this way, "you can have tens of thousands of users [on an internal network] and one public IP[v4] address", says Mr Rickard.
Some companies will need IPv6, he concedes, to make their network design more flexible. But most will live without it. "There is too much disruption for too little gain," he says.
Proponents of IPv6 argue that the technology solves several other problems. It enables greater security than the present version and should allow for faster data transmission.
However, detractors retort that these problems can be solved in other ways, with improvements in message encryption, security protocols and bandwidth.
While IPv6 remains a technically elegant means of achieving such things, the economic arguments for a costly move to the new protocol are less pressing for many companies. Companies that do hang on to IPv4 may even hold the advantage, in that they are impeding their rivals, who must install equipment that works with both protocols. Technological innovation carries a high price when the strongest force at work is inertia. |
