Academic journal article The McKinsey Quarterly

Time to Get Rid of Legacy Systems

Academic journal article The McKinsey Quarterly

Time to Get Rid of Legacy Systems

Article excerpt

The real issue is people, not systems

Here's how one utility addressed it

Everyone knows - or thinks they know - what the problem with IT is: large organizations competing in rapidly evolving markets are held hostage to the out-of-date computer technology they use to run their business. Year by year, the problem becomes more acute: the technology is that bit closer to collapse; additional product lines usually mean additional incompatible IT systems; company acquisitions bring along entire legacy systems of their own. Any new system must be capable of serving a wide range of needs; but because implementation typically takes years, the old systems must be kept up and running until the whole of the new system is in place.

Faced with this situation, most CEOs resist drastic action that may put at risk the operational stability of their company. "Evolution, not revolution" is the preferred tactic. Unfortunately, it often leads to simple inaction, which makes matters worse and saps competitive strength. And so IT becomes for today's company head a kind of bear confined to a cage that can't learn any new tricks, but, if fed enough - IT costs are now increasing at about 15 percent per year - will continue to perform its old ones.

An evolutionary approach is probably a competitive mistake on any terms. Most legacy systems are unable to cope with the date management problems associated with Year 2000 (brought about because early programmers coded years in two-digit fields to conserve scarce computer memory), and the costs and risks of making changes to vast amounts of unstructured and often undocumented code are horrendous. Other legacy changes driven by new legislation, such as the move to a common currency in Europe, pose similar problems. Some companies have estimated that almost all their IT resources will be consumed by these issues over the next three years, at huge cost but zero business benefit.

The good news is that replacing legacy systems is easier than it has ever been from the technology point of view. Proven off-the-shelf software now exists for almost every industry application. Where it does not, applications can be rapidly constructed from libraries of components - what is known as "object-oriented programming." The bad news, as companies about to bite the bullet in advance of the millennium are realizing, is that the technology issues, daunting though they are, have never been more than half the problem.

There is also a massive human resources challenge to face. The out-of-date systems are usually run by people with, to put it brutally, out-of-date skills. In a big company, there might easily be over 1,000 of them. Simple humanity as well as business interest behooves such a company to find a win-win solution to the problem. It cannot afford to stumble during the transition to the new system, and yet it finds itself crucially dependent on a large group of people who can see a system being put in place around them that may make many of them redundant, or require them to go through painful retraining or changes of job or location.

Faced with the urgent need to restructure its business in a short period of time, with legacy systems the main obstacle to change, one major utility recognized the criticality of the human element and developed an innovative approach. It employed a step-by-step business process that allowed both technical and human problems to be stated in a form comprehensible to general management and the workforce at large. With enhanced communication at the heart of the transition process, the organization completed one of the largest legacy migrations ever attempted - on time, on budget, and without seriously compromising the day-to-day operations of the business.

Its success stands as an example to those confronting legacy problems. CEOs should know that the bear can be released from its cage.

A bigger problem than usual

At the time it undertook the project, the utility provided services to 18 million customers in 12 regions via 12 major sets of processing systems, each with its own hardware and software. …

Author Advanced search


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.