Be afraid. Be very afraid. So goes a classic line from the Addams Family film. With the recent crash of the Internet in the US, it would seem the media was applying the same warning to the Internet. But is it really that bad? Can the whole Internet collapse just because one operator is a bit worse for wear from a night at the local cyberpub?
First things first. The Internet relies largely on the networks provided by telcos, many of which are gradually becoming telco/intranet companies. While the phone network provides excellent levels of reliability it is always subject to people drilling through cables, vandals destroying junction boxes and old-fashioned power losses. All these problems can have knock-on effects for your average Internet user.
But the recent US crash highlighted another problem for the Internet infrastructure. Namely that the company which maintains most of the domain name system, InterNic (part of NSI), has the responsibility for updating what are called root servers with update details which are sent as so-called zone files.
At 2.30am on a Thursday in July, an operator failed to notice a computer warning that the zone files were corrupt and allowed the files to be sent out. The result was that 35 per cent of US Web sites were inaccessible for over four hours. Millions of emails also failed to reach their destination.
To be fair to NSI, it has reviewed its procedures and claims that two people are now involved in the process that caused the problem, at least one being a senior manager.
But July was not a good month for US Internet users or providers. ISPs on the West Coast were hit when a local power failure knocked out service at Worldcom's MAE-West regional network access point. Almost a week later, the same system was affected when a circuit breaker failed. A day later, East Coast users of MCI, Sprint and Worldcom services were hit when construction crews cut into fibre cables.
The problem for many newcomers to the Net is not that it is falling down as often as a drunk getting home from an all-nighter, but that he/she doesn't know where the weaknesses in the system lie.
Few ISPs encourage subscribers to probe deeply into the reliability of their services even though the stats are usually good. Many subscribers are aware of the notion of modem ratios, but such a figure is of little interest until you know something about the volume of traffic modems are dealing with.
ISPs in the US are realising that part of the problem is that they need to work together better and ensure that rerouting of traffic is invariably an option when problems occur. Many leading US ISPs (including UUNet and Netcom) have joined together to form the IOPS.org Consortium which will look at ways of improving reliability, and consequently the image of the industry. The group is working on improving routing between ISPs, as well as establishing a contact at each ISP for use when problems occur.
For UK Web design companies changes can't come fast enough. "I don't want my customers to think the Net is crashing all the time, even though reliability is generally good," says Malcolm Greenaway, MD at Aquae Vitae Online. "The problem is one of how InterNic works now that it's becoming more business-oriented. You only have to look at other monopolies, like the provision of cement, to know it will lead to problems."
Concern about InterNic is echoed by Simon Murdoch, MD of online bookstore Bookpages: "Outage is a serious concern because it can always happen when you have an important customer about to place an order. On the whole, the Net is very reliable, but where one body like InterNic or Nominet controls domain names, there is always a potential risk."
It is the mix of commercial interests and technological innovation that is fuelling the whole debate about the ability of the Internet infrastructure to meet current traffic levels. Meanwhile, ISPs will capitalise on the concerns of business and stress the importance of using private networks to route as little as possible onto other telco/ISP services. By its very nature, the Internet is a global service with many weak links in the chain, but large telcos and emerging Internet telcos like UUNet will gain from the growing need of business to see a more reliable service.
Geoengineering on the sea floor near glaciers would form a new ice shelf to prevent melting
Alterations in capillary blood flow can be caused by body position change
Curiosity rover is in 'normal mode' but not transmitting scientific data back to base
NatWest outage comes a day after Barclays' IT systems shut out customers and staff