Thursday, 15 July 2010

G-Cloud – data centre or true cloud?

My latest project has given me considerable exposure to the design and development of new applications to be hosted in the ‘cloud’ - albeit for a commercial company rather than a public sector customer. However, it has led to a number of contacts asking for my opinion on the Government’s G-Cloud initiative.

On paper, G-Cloud is potentially a money-saving initiative even if, as I suspect, it is more a shared data centre than a true cloud computing initiative. The question is whether the project can be made attractive to the many, largely autonomous, organisations that form the UK Public Sector. Whilst Central Government departments can perhaps be expected to sign up to this initiative (although I doubt that there will be universal acceptance without the use of a stick or a very large carrot), past experience has shown that other organisations like local authorities, police and others will be far more reluctant.

As I’ve discovered in my cloud projects, once the security issue has been overcome, the next most important factor is cost and the ease with which additional computing power can be bought on stream to deal with peaks in demand. Can G-Cloud match commercial cloud providers such as Microsoft, Amazon or others?

My view is that it is unlikely to. Running a cloud purely for the UK Public Sector means that it will have to be sized to cope with the peak demands of its customers – peaks that will in many cases all occur at the same time. If a flexible pricing policy is adopted, then I suspect that it cannot be competitive with commercial cloud suppliers who manage a wide variety of peak demands, and can therefore spread their costs better. And what comes first? – the computing power or the demand? – in a public sector heavily constrained by budget restrictions can the computing power be put in place before the demand is contracted?

No – I suspect that G-Cloud will be more a shared data centre, with organisations committing to take up a dedicated level of computing power, with the level of pay-for-what-you-use computing limited by relatively high on-demand pricing (but still likely to be cheaper than running in-house options).

Also, it will be interesting to see if the concept of the Government App Store succeeds or not. In theory it should, but in practice I fear it will be limited by the current architecture of many of the existing applications used within the UK Public Sector. In the short-term, what I expect instead is a few suppliers cleverly offering a SaaS pricing approach on existing ‘legacy’ applications without embracing a true on-demand use of hardware – most probably against a ‘minimum commitment’ that will limit the cost savings for users.

But given a software generation or two, I believe new applications will be developed that have been designed to make optimum use of true on-demand, cloud computing systems. Only then will the real cost benefits of cloud computing possibly be realised by the UK Public Sector.

No comments: