Information systems infrastructures are close to collapsing through complexity.
They're like airports in international business hubs such as New York, Frankfurt or Tokyo, which were never designed to handle today's air traffic volume, air traffic control complexities, security requirements or parking needs. As a result, they're now in a constant state of frenzied renovation. They are, alas, legacy infrastructures.
That's also true of just about every large organisation's IT resources. We're living with enterprise technology platforms that are an ever-more-complicated lashing together of designs, devices, standards, servers, operating systems, database management systems, legacy systems and application packages.
As with airports, companies will have to live with much of that heritage for an unknown period. They will add on and patch in new components. They will grab any tool, such as Internet technology for creating intranets, that reduces incompatibility and integration.
But at some stage, the sheer complexity of the IS technology base will overload the IS management and skills base. Already, from the perspective of most users, we're gaining only marginal improvements in value from new IT tools relative to their growing costs, dependence on overburdened support functions, administrative procedures, breakdowns and unreliability.
The CIO of a leading petrochemical company told me a few weeks ago that all the talk about IS strategy is a facade. His entire focus is on keeping the nearly 100,000 desktops in his company up and running. He's barely able to do that today, and he dreads the impact of laptops, networked multimedia, electronic commerce, intranets/extranets, agents and wireless communications on his staff's ability to keep up with demands.
For this CIO and his peers in other large companies, it would be as hard to put in place an entirely new infrastructure as it would be to re-engineer JFK Airport. The ideal answer is to tear down JFK and start again, but that isn't practical. Neither is tearing down the IS infrastructure. Legacy systems get passed on from one generation of business, technology and applications to the next.
That said, surely it's time to start thinking ahead about the target architecture for IS and then to choose the tools that maximise the chance of being able to break away from the current legacy/heritage design constraints.
Here's my own assessment of the necessary principles for defining next-generation infrastructures. They must meet the following criteria:
-- Be designed from the customer back to the company and not be based on distributing capabilities from the company out to the customer as is currently done. That probably means that the lowest level of standard Web browser will be the basis for design because that maximises the customer base. (Yes, that does mean giving up a lot of features such as Virtual Reality Modeling Language and much of multimedia for the time being.)-- Ensure complete simplicity at the customer end. That simplicity must encompass everything: learning, troubleshooting, installation and access to support. Windows 95 and the hardware/software industry that has grown up around it are a disgrace in this regard. They put far too much complexity on the user and take far too little responsibility for resolving the problems that complexity creates on a daily basis.
-- Minimise variety. A heresy perhaps, but it's close to impossible now to manage the desktop software environment plus the plethora of data warehousing and data mart tools. I see no choice but for companies to strip down software and hardware options, not in the name of standards but in the interest of the sheer survival and sanity of IS staffers and the user community. In networking, I suspect that means standardising on IP.
-- Reduce learning and support requirements at all levels. By reducing the need for users and IS to learn technology, there will be less need for user support by IS and IS support by vendors.
-- Be commitment- and promise-based, not feature-based. The promises made by most IS organisations are about as reliable as the airlines: "We'll try to get you there on time. We may have to cancel the flight. Well, things do break down, you know." The IS architecture of tomorrow should be based on firm promises and commitments: "If we install it, it works. We guarantee the quality of the software we offer."
-- Outsource in the interests of simplification of skills needs, management attention, support and architectural complexity - even if it increases immediate costs.
Most of the companies I work with are at the opposite extreme of each of those principles for simplification. Their IS units are working ever harder to add complexity. Let's stop. The policy for IS should be simplicity, not high-tech. If IS remains a creator of complexity, CEOs will find a simple solution: Outsource the lot.
(Keen's book The Business Internet and Intranets was published last month by Harvard Business School Press. He can be contacted at email@example.com.)