The government's new Cloudstore, an online catalogue of cloud-hosted IT services for use by public sector organisations that was launched yesterday, is an unusual combination: it is a government IT project, and therefore subject to public scrutiny, and yet it is experimental, even by private sector standards.
Most of us in the IT industry know the arguments for cloud computing by now – consuming IT as hosted, scalable services should improve utilisation, cost transparency and organisational agility.
But few organisations, certainly none the size of the UK public sector, have taken this App-store like approach, that allows divisions and departments to pick cloud services from a shortlist of approved, competing suppliers.
Chris Chant, programme director for the G-Cloud, makes a strong case for the model.
"Using cloud solutions that have already been secured and accredited will almost always be cheaper," he said during a talk at the Institute of Government last year. "We'll know from the outset the cost of the product, and most importantly we'll know the cost of exit."
"And price and quality will be visible to all," he said. Having a review on each service from, for example, the CIO of the department that used it would "transform pricing and quality, far and away above what any service level agreement has ever given us".
The combination of G-Cloud and the CloudStore, the government predicts, will save £20 million in the next fiscal year, £40 million the following year, and £120 million in 2014-15.
But as I'm sure Chant is well aware, getting the website up and running is just the first step in a long process, in which many challenges will need to be addressed and many questions answered.
Firstly, how will cross-departmental cloud procurement be governed in order to guarantee cost reductions?
Speaking at the recent CloudCamp event in London, Joe Weinman, head of HP's media business and author of Cloudonomics, explained that the main reason why cloud computing can save you money is that you don't pay for it when you're not using it.
"Cloud computing isn't necessarily cheaper to use," he said. "But it's free not to use."
This suggests there is a risk that cloud computing can end up being more expensive than in-house or traditionally hosted alternatives, if it is used for systems that are in perpetual use. Are the governance mechanisms in place to make sure this doesn't happen?
A related matter is the argument put forward by Simon Wardley, of CSC's Leading Edge Forum. Referencing 'Jevon's Paradox', Wardley argues that if cloud computing were to make IT services cheaper, organisations would find more ways to use them and therefore both consumption and expenditure would increase.
This is not a problem, as long as you believe that in consuming more IT services the government will necessarily deliver more value to taxpayers. But surely there is a danger of 'cloud sprawl', the ease of provisioning new infrastructure leading to unnecessary expenditure.
The government's CloudStore is actually more of a market, allowing buyers to compare and contrast similar services from competing suppliers.
The ability of this market to function properly relies to some degree on commoditisation of certain IT services. It is entirely appropriate to compare suppliers offering the same software on the same hardware on the basis of price and customer satisfaction.
But the store is not limited to infrastructure-as-a-service. It also includes platform-as-a-service offerings such as Google Apps, and software-as-a-service applications such as SAP's BI OnDemand, offered via Atos.
Is it fair – or more importantly, useful – to compare these kinds of system, whose utility depends mostly on how you use them, on these terms? Is it helpful to imply that selecting a business intelligence tool is as simple as choosing a content delivery network? I suggest not.
And finally, how much of the government is in a position to provision its own cloud-based services via the CloudStore? The government is locked into a number of multi-billion contracts with suppliers who own their own data centres, and are unlikely to let their government clients switch to a rival. Maybe this explains the savings targets which, in the context of UK government IT, are miniscule.
The reason why I raise these issues is not to give Chant and the G-Cloud programme a hard time. To the contrary – it is to their credit that they have launched a project that touches on such cutting edge IT management concerns in the context of the public sector.
But in doing so they have produced an experimental test-bed for an emerging IT procurement technique, which is why Information Age will be keeping a close eye on the fortunes of the CloudStore.