Original URL: http://www.theregister.co.uk/2006/01/18/utility_data_centre/

I'm dreaming of a utility data centre

Nothing but a dream?

By Martin Banks

Posted in Servers, 18th January 2006 12:40 GMT

Only time will tell just how much of business IT’s future they will come to represent, but Utility Data Centres (UDC) are bound to play a significant part in most companies’ IT infrastructure plans over the coming years.

The "Utility" part of what is otherwise a less-than snappy hook on which to hang a future is meant to convey the idea of information processing being available in a way analogous to the utility services of gas, electricity and water provision: turn on the tap and there it is (well, there it usually is). It is arguably the "great dream" that the major IT vendors – IBM, HP and Sun primarily – are trying entice users towards, though to be fair those companies are still largely groping their way to what they hope will be a future business.

Utility, ITility

The UDC idea brings together many of the trends now prevalent in IT – increasing use of standards in applications development, the growth of the standard platform, the move towards Blade servers and the growing use of virtualisation technologies – to postulate a platform which is ideally suited to that other great dream, the shift to a pay-per-use, service-based infrastructure built on the concept of the Service Oriented Architecture (SOA).

Just how deeply the UDC trend will embed itself in the hearts and minds of users is hard to tell, for there is still considerable debate about whether it is better to centralise IT resources as much as possible, or distribute them widely. For most businesses, the answer will be the classic compromise of "a bit of both", with widespread use of small, powerful personal clients - PDAs-on-steroids – but with as much of the back-end infrastructure, held under as close control as possible, in a datacentre. If nothing else, the growing demands of compliance and governance are likely to push many users towards IT centralisation whether they like it or not.

So the UDC idea will probably start life by finding its way into many larger enterprises, particularly as businesses develop processes built as services rather than based on specific applications. It will be an intra-company function, using racks of Blade servers running in a virtualised environment to provide internal users with a resource pool of the hardware, operating systems, applications and tools needed to build and initiate a service, run it, and tear it down cleanly when the task is complete.

For many users this will be as far as they ever get down the UDC road, but there are ways in which it can be developed that exploit the pay-per-use model by taking it into an open marketplace environment. For example, some users may realise that they have services that are marketable to others. But rather than sell a licence to the code, a revenue stream can be created for the IT department that built them by providing them as services run on the service provider’s existing datacentre.

The next step down this path would see the development of Service Aggregators, companies that bring together services from a range of service providers – both hardware and software - that meet customer needs. The chances are such aggregators will start life specialising in definable market sectors where they already have the expertise and the experience needed to identify the most appropriate services and tools for a market sector. Major IT vendors such as HP are already eyeing this potential market with some relish.

Tapping the technology

Long term, or so the theory goes, the majority of users will gradually shift towards utility-based service provision for all their IT needs. They will specify their basic resource and service requirements and be able to add short-term services and resources, at a cost, when required. This will be particularly important for small companies, for they would be able to pay as they use the platform resources and applications licences that otherwise they would not be able to afford to buy outright. That way, they could compete on level terms with much bigger rivals. And for any company aiming at the maximum in agility to meet changing market demands they could, so long as they could pay for its use, simply "open the tap" and there the necessary resources would be. A marketing campaign that generates more business than even the most wildly optimistic marketroid’s forecasts? Just open the tap and get the resources you need.

Is that final scenario a dream? Possibly, and it is almost certainly true that it will take longer to appear than many of the vendors hope and expect. But how many people can remember seeing factories with their own electricity generating stations, and how many factories are there like that now? How long before "IT" stops being a core interest to a business and instead the attention turns to what can be done with it for the good of the business. Then, it may not matter from whence the "IT" actually came.®