The death of the data centre?

19 Dec 2011

Rumours of the demise of in-house computing have been greatly exaggerated, says Clive Longbottom, service director at market analyst firm Quocirca, in an extract from his IP EXPO 2011 Pathfinder session.

The ‘death of the data centre’ has been widely discussed in the press. I don’t agree with these predictions. To me, they’re like those ‘death of the mainframe’ stories, which have been appearing for 30 years now, even though IBM still sells more MIPS [mainframe instructions per second] each year than it has ever done before.

In February 2011, the team at Quocirca conducted some research on behalf of Oracle, looking at the next-generation data centre. We talked to 900 large businesses about their perception of the cloud, and the results showed that cloud is not as far forward as the media and some analysts would have you believe.

In that survey, over 20 percent had no plans whatsoever to support cloud and a further 13 percent said they didn’t know what their plans are. Some had plans to use only private cloud computing, others are sticking to public cloud services only. A very low proportion – around 7 percent – said they plan to use a hybrid mix of internal/external private clouds, mixed with public cloud services as required.

This hybrid model is the way we at Quocirca see things going – but we also see a great deal of confusion that is likely to impact on how companies move to optimising their own data centres.

Yes, the widely touted benefits of public cloud services are attractive. Yes, IT is a mess in many organisations. But these, in themselves, are not good enough reasons to move away from the internal data centre.

It’s more likely that, at most organisations, some workloads will move to public cloud services, while others move into a private cloud, either within a fully-owned data centre or a colocation facility. In order for that strategy to be successful, a cloud strategy will be needed.

1.    Understand what you’ve already got. Measure it. Monitor it. Manage it.
2.    Assess the suitability of applications and services for cloud migration. Which can be moved to a cloud environment? Which ones can be sliced and diced, with some of their functions being moved to the cloud? Which ones are better provided by an outside provider? What are the business risks associated with moving these apps to the cloud? What are the potential business benefits.
3.    Recognise that new functionality demands planning. Any changes will need careful consideration of the relative merits of on-premise or cloud provision.
4.    Map out what you’re going to move, how you’re going to move it and when. Above all, have a ‘Plan B’ – if things go wrong, you need to be able to ‘roll back’ applications from the cloud.

The internal data centre is alive and well, but it has to change dramatically. It may get smaller in the years ahead, but it will remain core. Most companies will use a mix of private and public data centres, but existing approaches will continue to be woven into the overall corporate IT strategy. Colocation centres that provide highly virtualized private clouds to corporate customers are booming, but this is just moving equipment. Those customers still need to own the hardware and they still need to optimise the environment.

The argument isn’t with cloud computing itself. Cloud will happen. Cloud is a big changer in the market.

The question for IT teams is which applications should be moved to the cloud immediately, which should move in that direction over time and which should stay within the physical environment for the time being.

Existing investments have to be sweated. It’s what we call a ‘career-limiting decision’ to go into the main board and say: “Hey guys, we’ve heard a lot of stuff about cloud, we’ve read a lot of stuff about cloud and we think we should move to a cloud environment. We’re going to get rid of all our Sun servers, all our AIX servers, our mainframe and replace it entirely with Wintel.” The board will simply turn around and say: “Here’s your P45.” So make sure you get the most out of existing assets – but only where they are fully supporting the business.

The video of Clive Longbottom’s full presentation is available here