Software13.06.2013

A DIY cloud is no cloud at all

By Alexander Mehlhorn, CEO, Framework One

There’s a great deal of pride and enjoyment that the geeks in an organisation get from building servers from scratch, creating server rooms, and architecting an infrastructure that they can boast about to their peers. And five years ago, this was an essential in-house skill. You could contract the creation of your IT infrastructure out, but that was costly, and risky too. After all, isn’t the IT infrastructure the backbone of any modern large organisation – and can you really trust someone else with it?

Companies – and their resident geeks – have become accustomed to building it themselves, investing millions in state-of-the-art data centers, often right in their head office premises. They’re used to having their IT literally at their fingertips. It was great for the geeks, but a nightmare for the CTO when it came to operations, scaling, staffing, support, security, and the host of other concerns that ensures CTOs never have a good night’s sleep.

And then it all changed. The biggest IT companies in the world realised that they had capacity in terms of hardware and skills that was underutilised. At the same time, the internet achieved bandwidth and price points that made it viable to move large portions of a business’ infrastructure and services to what we now call the cloud.

It makes perfect business and architectural sense. Real hardware is clunky and expensive. It takes up space, generates heat, and consumes power. Building redundancy worthy of the name is expensive, involving geographically dispersed data centers. The networking requirements are complex. The security – both physical and virtual – requires constant vigilance. And it really is not the core skillset of most companies. What you want is a company like Microsoft that does IT alone taking up this complexity, and delivering a cloud solution that is simple to manage, and lets you get on with your company’s core competency.

But it’s been a very sudden change. For many of the geeks, the CTOs and the businesses that enjoyed the comfort of being able to walk into their own server room and feel that sense of ownership, the change has been too sudden. There’s a certain level of fear and mistrust when it comes to cloud computing. There’s a feeling of loss of control. For some it’s a fear of losing their jobs as their technical expertise in running an old-fashioned data center face obsoleteness.

Caught between the obvious advantages of cloud computing and this fear, some companies are making the critical mistake of attempting to build their own cloud solutions in their own data centers.

Trying to do so negates all of the benefits of true cloud computing. You’re still burdened with the crushing complexity of running a full datacenter, and the company will continue to operate like an IT shop even when its core business is something completely different. But mostly, you’ll lose out on the mind-numbingly massive scale that you can achieve with cloud computing.

If you think your datacenter is big, imagine hundreds of those stacked end-to-end, replicated across six centers, on three continents, like Microsoft’s Azure cloud computing platform. Microsoft uses containers filled with servers to build its data center. Each container contains between 1,800 and 2,500 servers, stacked two high, and a new container can be added to the datacenter in under a day.

Companies like Boeing and Coca-Cola have turned to Azure, as they simply can’t achieve the type of scale that Microsoft can with its cloud infrastructure. Cloud computing gives these companies the flexibility to be competitive, and that’s the real benefit of cloud computing. Reduced cost and complexity is great, but being able to respond in real-time to business demands on IT infrastructure, and the competitive advantage that that brings to a company – that’s what being in the cloud is really about.

Sign up to the MyBroadband newsletter