As we enter 2005 , IT leaders are once more examining spending priorities to determine IT investment for the coming year. According to Gartner "CIO’s and other IT executives will have to spend the next 12 months holding down costs while innovating for the future." This is the CIO’s dilemma for 2004/5 - to achieve a balance between managing short-term issues of cost and complexity, while at the same time investing in the future. One of the key items for the CIO in 2005 is to decide which legacy systems should be replaced and which should have their life extended. This shouldn’t come as a surprise as legacy systems play such a critical role in IT infrastructure; yet seem to drain precious resources.

This article explores the business dilemma for the CIO, namely to hold down costs, innovate for an agile future, and decide what to do with legacy systems. Each taken on its own is a formidable challenge. Can there be a way to deal with all three at once? To answer this, we have to turn the world over, to look at things from an unconventional perspective…

Conventional wisdom tells us that new is better than old. New hardware is better, cheaper and faster than ever before. Replacement of old computers by new computers is necessary for organizations to deliver products and services better, cheaper and faster than the competition. Consumers who have come to expect year-on-year improvements fuel this cycle. Likewise, the Internet is better than old proprietary networks. Windows XP is richer in function than DOS and Linux is cheaper than proprietary mainframe operating systems.

While legacy hardware may need replacing, we must then ask what is to be done about the applications running on that platform. After all, it is the applications that fuel the business, touch customers and help deliver products and services to new markets. Does all this mean that new applications are better than those written in previous decades? Are COBOL business processes, the most widely used in the world, obsolete simply because they can be re-written in Java? Probably not.

The business procedures written down and captured, whether in computer language or English (or any other spoken language), may be relevant for many decades, or even for centuries. For example, the City of Massri Treasurers used double-entry accounting (implemented by all major financial packages today) as early as 1340 A.D.