The Millennium Bug: Is your system critical, peripheral or in a grey area?

With time running out for making computer systems able to cope with the year 2000, many companies are being forced to reassess their entire IT strategies. Stephen Pritchard reports.

The idea of the Millennium Bug as a disaster on a grand scale has gripped the public imagination, with predictions of the wholesale failure of anything from traffic lights to air traffic control systems.

The reality is likely to be less dramatic. Some systems are bound to fail, but in most cases industry analysts believe that even if it is too late to fix the bug itself, there still time to organise contingency plans.

Experts on the problem divide computer systems into three classes. As Tony Hill, UK general manager of Intersolv, explains, there are critical systems businesses must fix, peripheral systems that there is little point in fixing, and a large grey area in between that should be fixed if time and resources permit. If there are failures, it will be in this middle ground: support systems such as databases and non-mission critical financial software that businesses find useful, but could probably live without.

As Hill suggests, any solution has to fit in with a business's financial circumstances. "You have to look at a cost-benefit analysis. For many companies, it is probably the first time they have had to step back and look at all their systems."

Faced with cash requests from the IT department, a finance director would be on solid ground in rejecting all but the most essential upgrades. Companies worry about the wisdom of spending money and time on systems that may, Millennium Bug or not, have only a limited shelf life.

Intersolv applied this logic to its own, in-house accounting system. "In our own case, we moved to another accountancy package sooner than we might otherwise have done," Hill says.

This sort of reaction is not always negative. Pressure of time limits the amount of creativity IT professionals can apply to the problem. "What we see happening is, companies either look for a fix, or just replace the system," Hill says.

The problem can also be turned on its head, and seen as an opportunity. Some millennium tasks, such as auditing IT systems, are valuable in their own right. Looking again at older computer systems can suggest ways to apply new technologies to them, and make them more useful. This approach is advocated by Keith Ireland, head of the professional services division at the IT consultancy and programming house Micro Focus. "People are having to revisit systems and think about making them compliant," he says. "Some people take the view that they can throw a system away and start afresh."

The real value of a computer system is more likely to be the data than in the hardware. Companies are finding, through audits, that they have considerable investments tied up in older computers, and there are ways to extend their useful working lives. Computers can be given a second life by being connected them to a local area network, or even to an intranet, so that anyone in the organisation with a Web browser can use the data. This, Ireland believes, is also a way to persuade the board that the spending is worthwhile. If the system has to be upgraded anyway, a new interface need not add significant extra cost.

"All you need to do is front-end the system with a technology that is user-friendly, and allow access to the legacy systems in a way that 'future- proofs' them. We have technologies that allow us to build a Web front end for systems that companies already have."

Recently, Micro Focus developed a Windows graphical interface for the ports company Southampton Container Terminals, which runs its container tracking system on an ICL mainframe, with software written in Cobol.

A graphical interface alone does not make a computer millennium-compliant. As Keith Ireland stresses, "you still have to fix the problem". This does not necessarily mean rewriting data. The core of the millennium problem lies in the way we write dates. Computer programmers of 20 or 30 years ago saved time, and valuable memory space, by shortening dates so that 1998 became 98. To a non-millennium-aware computer, that year could be 1998, or 2098.

"The obvious thing to do is expand the date - for example, to make it four characters," says Tony Hill at Intersolv. "In practice, many people are not doing that, but making changes to the software that accesses the data."

The technique, known as "windowing", makes assumptions about the likely real date when it is presented, either by the old computer system itself or to an interface on a PC. A program simply makes the arbitrary choice that 1/3/98 is more likely to be 1998 than 2098. "It is a work-around that may last for 10 or 15 years, which buys them time to fix it properly," Hill says.

The windowing approach fits in well with the idea of using new interfaces: the job can be done at the same time, often with the same software. A more radical alternative is to ask whether the program is needed at all. Sometimes, the best solution is to find a completely different way to carry out the task. "Sometimes, people re-engineer the business processes the computer system supports," says Tony Hill.

Finding a totally new and millennium-compliant way to run part of a business system avoids the potential errors of reprogramming, freeing up time for more pressing year 2000 concerns - such as a cheap supply of good champagne.