Devil's Advocate: Ignore IT change at your peril

So many IT projects seem to go wrong. But, argues Martin Brampton, factoring in incremental pressures on infrastructure is the small thing that can make a big difference.

A lot of IT people are really quite good at what they do. Yet the cry forever goes up that top management is not getting value for its IT cash. A big part of the problem is in the management of infrastructure costs.

Projects that go off the rails with huge price tags inevitably hit the headlines. But it is worth repeating that nobody else has the answer to managing large, complex projects. Concorde was spectacularly over budget, despite being almost entirely an engineering project. Most military hardware comes in at vastly inflated costs. And so on.

But the insidious element that frequently undermines expectations in IT is infrastructure. As with other specialisms, there are plenty of skilled infrastructure managers. Given the materials they work with, they have often been remarkably successful.

The comparison has long been made between basic computer services and the office telephone system. People just take it for granted that if you pick up a telephone, there will be a dial tone and calls will connect. Computer services have been getting closer and closer to that ideal, as more and more workers sit in front of a computer screen.

Quite apart from the demand for a high-quality service, there is also a lack of patience with reasons given for IT failures. Infrastructure problems are often extremely complex, related to many separate factors, and inherently hard to explain to the impatient user. This is true whether the user is the office junior or the chief executive.

Moreover, IT infrastructure is constantly changing. Implementation of new facilities is often costed so that it seems that each innovation should bring substantial financial returns. Yet what is nearly always ignored is the incremental effect on the general infrastructure. Simply leaving basic systems such as networks unaltered for a period is likely to lead to a situation requiring heavy investment for no apparent return.

Vendors are often little help. Software is written to appeal by reference to its features and implemented to be as simple as possible for the vendor. Rarely is attention given to the costs behind the scenes, an area left to the patronisingly named 'techies'.

To take one simple example, with bandwidth now relatively plentiful, there are very strong arguments for running software on servers with workstations only providing standardised front ends. Increasing reliance on standalone portables undermines this fact. But more than that, vendors who consider server-based operation a luxury create problems.

Occasionally a determined effort is made by skilled IT staff, with the backing of an insightful management, and a wholly centralised system is achieved. Even then, it is difficult to maintain this structure against the constant demand from sections of the organisation for whatever new application seems to meet a need.

The alternative approach to keeping user facing machines under control is automated software distribution. For a setup that has not yet achieved this, there are major obstacles to achieving the base position where automated distribution can begin. Worse, when achieved, the automated system is fragile and neglect can quickly lead to a situation where it ceases to be effective.

If managements want better value from IT systems, they will have to make more effort to factor in the incremental pressures on underlying systems that come from new developments. And infrastructure managers will have to find ways to give clear explanations of the complex issues that they face in maintaining a high quality of service. Until then, the best that IT managers can expect is to be ignored.