Most people now agree the client/server approach failed, and Forrester Research has a new model: organic IT. I recently met Forrester Research analyst Frank Gillett, who has been a driving force behind the new concept.
Gillett's first remark was that client/server computing is inefficient. For instance, programming with Java or the Microsoft application programming interfaces leads to proprietary code, which increases labour costs. The answer to this dilemma is XML web services, which provide a standard way to build applications.
His other main bugbear was that client/server computing is remarkably inefficient in the way it uses computing resources.
The classic example is a website that needs to cope with a peak of 100,000 users on the last day of a marketing promotion. To support this peak, the IT director needs to ensure the datacentre for the website has enough capacity in terms of processing power, network bandwidth and storage to cope with the huge influx of users. For most of the time this expensive computing infrastructure sits idle. Gillett argued that during the idle time, the IT resources could be redeployed, and thus improve the utilisation of hardware.
Up to a point, this makes perfect sense. Why should I need to buy 2Tbytes of storage up front when I only need 500Gbytes today. It is cheaper to buy the lower-capacity disc system and expand as and when I need to. As a bonus, I may find that the cost per terabyte of disc capacity falls over time. The same is true of memory or processing power. This is simply the application of Moore's Law: performance doubles every 18 months while the cost remains static.
This is clearly great news for users. If organic IT succeeds, you will never need to buy expensive hardware up front, instead waiting until you need the extra power or headroom it offers.
Somehow, I don't think this is how the industry sees organic IT, as I believe it could restrict how hardware firms position their enterprise hardware. "Room for growth" and "headroom" are simply other ways of referring to under-used and expensive IT infrastructure.
Suppliers need users to buy this hardware to recoup the cost of the research and development work they have undertaken in creating such products. Rest assured, they will make their money back somehow.
This was first published in May 2003