In 1943 Thomas Watson, IBM’s founder and president, assessed the commercial potential of ENIAC, the world’s first electronic computer. “I think,” Watson is reputed to have said, “there is a world market for maybe five computers.” Today, Watson’s remark is usually cited as an example of how even the best-informed people can vastly underestimate the potential of new technology.
But actually, Watson had overestimated demand: IBM only sold two of the systems. The problem was that these 30-ton machines could only be used to perform one specific task.
In the 60 years since, the computer industry has continued to struggle with designing systems that solve a wider range of real business problems. It has been a painstaking journey that has been only moderately successful.
It began in earnest in the 1960s when IBM’s System/360 introduced the concept of systems architecture for the first time. It marked the recognition that since business problems come in all types and sizes, computer systems need to be flexible and scalable to address them.
Since then there has been the time-sharing mini-computer systems of the 1970s, intended to make departmental computing economical, and early network systems that allowed IT groups of workers to share computing resources.
In the 1980s microcomputers addressed personal productivity issues for the first time. Subsequently, the Internet and wireless technologies have connected giant corporate data centres to palm-sized machines in the pockets of users, extending access to business systems.
Systems architectures continue to evolve, since there remains an issue of how to build systems that can quickly adapt to the changing requirements of business. For all the developments in architectures in the last 60 years, each new improvement has required organisations to compromise – adapting their requirements in some way to match the imperfect capabilities of the systems available to them.
But perhaps the industry is getting closer to delivering that perfect systems architecture. Service-oriented architecture (SOA) promises to accelerate our ability to build and deploy new applications; grid and autonomic computing may offer the key to dynamically allocate and reallocate computing resources to changing needs and workloads. We may be on the cusp of a golden era, where organisations will utilise event-driven architecture (EDA) systems that respond automatically to shifting requirements – in real-time.
But here’s the twist: by then organisations will not necessarily own any computer systems at all, but will access all their IT needs from a few – perhaps as few as a handful – of utility computing service vendors. In effect that will mark the death of the IT architecture as we know it. As Thomas Watson might have observed, predicting the future is an imperfect science, and perhaps the only thing that is certain is that when the perfect systems architecture arrives, systems architecture will no longer be a critically important factor in business strategy.