Cloud computing is everywhere right now - and, unfortunately, everyone is using the term quite loosely. The biggest feedback I get from CIOs is a single instruction: "Don't use that term because we need something more tangible". So let's try.

Analyst firm Gartner recently predicted that, through 2012 at least, enterprises will spend more on private clouds than public cloud service providers, so it's worth knowing more about what constitutes ‘private' and ‘public' in this sense. Put simply, private clouds are similarly elastic to external services and are delivered over the internet - but they are delivered by IT departments to internal customers only. Public clouds, on the other hand, are typically run by third-party providers for a range of customers.

As Phil Dawson of Gartner states: "With cloud offerings coming in the form of services, this means that the IT organisation will be replaced by relationships to many cloud computing service providers, each for one or a handful of services. Larger organisations will continue to have an IT organisation that manages and deploys IT resources internally. Some of these will be private clouds, but not all. IT departments will also take on IT service sourcing responsibility - determining when to leverage external providers, when to deploy internally, and when to leverage both for specific services."

As Dawson suggests, the picture will be mixed for some time with some though not all using private clouds as a stage towards public versions of the same. Interestingly, far from viewing it as a panacea, Gartner also sees limited application of private clouds with over 75 per cent of use focused on "very large data queries, short-term massively parallel workloads, or IT use by start-ups with little to no IT infrastructure".

So, far from being some amorphous, woolly concept, cloud computing starts to become a little bit clearer once we define our terms and accept that some purposes will fit it neatly for some companies - and others perhaps not at all.

I tend to see the private cloud as part of a journey or a multi-step process. Different companies will pick up key services for the cloud but adoption depends on the characteristics of those firms, their markets and their appetite for risk versus control.

Once firms start using the cloud, what will really differentiate the success of cloud computing will be just how ‘virtual' datacentres can become. It will be the automation and management side that will differentiate ‘good' clouds from ‘bad'.

There will be quick wins to be had in basic automation and ringfenced services but the biggest successes will come from fully virtualised infrastructures that use a hypervisor and cloud operating systems with service levels applied. Storage and networks will have to be unified so that you have total governance over the information flow and can apply policies and become genuinely service-oriented.

But most people aren't considering the challenges of management, and there's a lot of scepticism about cloud service providers because there will need to be high levels of trust. However, familiarity will also be a strength for companies that come from a consumer background like and Google.

There is a lot of talk about the opportunity to charge back for cloud services and introduce pay-as-you-go pricing, but that is probably step two when you can manage your utilisation much better and have what some people are calling ‘infrastructure as a service'.

But the biggest challenge is not the technology but the ability to change at speed, because there are significant challenges in terms of collaboration. You're combining a number of silos into a single provisioning system and crossing a lot of stakeholders who ‘own' various services.

Technologically, however, things are moving fast. In small and mid-sized companies with fewer than 500 servers, 70 per cent of servers are virtualised. In large enterprises with 1000-plus servers it's quite a different picture, with significant legacy meaning only 20 to 25 per cent of servers are virtualised and firms are struggling to accelerate. Sometimes the transformation programmes that are underway are too large and they've taken on technological progress without business buy-in.

The biggest problem after change management is cost, but if you can force through change, cost becomes a real reason for doing cloud computing. With single-click provisioning and higher utilisation of servers, storage and network ports there are 50 per cent capex savings and 40 per cent operations savings to be had. But you've got to cut the elephant up to eat it and that's why we should expect parallel programmes between private and public clouds for some time to come.

Read Mike's column on the Microsoft cloud strategy