Recent developments in ICT have made it possible to share ICT infrastructure in a way that delivers increased flexibility and responsiveness to business needs whilst reducing costs. This change involves a move from ICT that has been procured separately by organisations as ICT infrastructure, to a new model in which ICT is provided as a utility which is known as “cloud computing”. This has been likened to the changes in the electricity industry during the early part of the 20th century as organisations moved from buying their own generators to procuring electricity as a utility.
The term cloud computing was originated by the large Internet firms where the rapid change and growth in their businesses provided the driver for developing a new approach to the provision of ICT infrastructure; an approach where the provision of standard ICT services needed to support customer facing activities was separated from the detail of the computer systems in use and their physical locations. As the sophistication of this approach developed, these businesses became able to regard ICT infrastructure as a pool (or cloud) of infrastructure resources.
As well as enabling business flexibility, the cloud approach also provided other benefits. Specifically:
- Standards evolved that made it possible the need to be able to deploy business applications on any available computer systems, rather than just those that had been uniquely configured
- Unit costs of computer resources fell substantially as the flexibility of allocating workload to any available computer system enabled much higher system utilisation levels to be achieved
The cloud model enables further significant cost savings over and above those from data centre rationalisation – the additional benefits arise from the flexible allocation of computer resources to workload and the cloud model also enables further reduction in energy consumption and much greater flexibility.
While the cloud model is sufficiently proven for there to be clear benefits to the public sector, it is still early days. The main challenges to overcome include confidence in information assurance, achieving guaranteed service levels, and determining the standards to adopt. It is clear however that there will be a major shift in the ICT industry to the cloud model, and that the benefits will be substantial. In the relatively short term, it will be possible to mitigate many of the risks through putting in place a private cloud for government.
Our vision is to develop and implement a government cloud infrastructure that enables public bodies to source ICT infrastructure, development capabilities and software applications from a secure, resilient, flexible and cost-effective service based environment.
Work on developing the G-Cloud strategy will be initiated in the autumn 2009, and will target delivery of a prototype development infrastructure in early 2010. This will be used to validate the model for a number of initial services -with goal of having the new approach ready to use as the standard model by the end of 2010.
Establishing the G-Cloud will involve a major change in the way that ICT is procured and supplied which will require significant change both in ICT suppliers and public sector organisations. Cloud commercial and business models are in their infancy and we will need to support the industry in developing the business case for investing in this new model.
Development of the G-Cloud is the key enabler of the £1.6 billion savings from ICT outlined in the Operational Efficiency Programme report (the remaining amount comes from collaborative procurement of ICT goods and services).
Run to the hills! Somebody has told the IT-illiterates who drafted this a new buzzword! The Cloud!
This will not end well. Fortunately for us, they seem not to understand (a) what ‘The Cloud’ is, and (b) why it’s utterly inappropriate for government data.
Like the introduction, this entire section is another of those “I am an IT illiterate” t-shirts worn by whoever drafted this drivel. The entire page should be ripped out, burned, and the ashes buried.
Note to all IT illiterates: government departments sharing servers and databases is not “the cloud”. The cloud is about users not knowing nor caring where their information is stored.
This is INAPPROPRIATE and UNACCEPTABLE for government data.
I totally agree with your comments. This is the biggest load of rubbish I have read in a long time. I would be scared if I ever thought these ideas would ever come to fruition. Just trying to imagine the scale of the projects makes me go cold and I have been involved in some pretty big projects. I think we can safely say that any projects based on this drivel will never see the light of day!
This seems to be about centralisation and virtualisation of datacenters – not about the INTERNET cloud your commentors are discussing (indeed the report talks about a government/private cloud – something Amazon, Google and Nicholas Carr acknowledge is necessary in certain cases -like this). In particular why should all the government departments run their own mail-servers paying civil servants to maintain small data-centres for that purpose. Better to virtualise then run them at highly efficient data-centers shared between government (but with virtualised walls seperating them). That way (as a guess) the Met office might be able to do its analysis at night using the capacity availble from interactive servicves which only run during the day (e.g. DVLA online payments)…
Come now, government and its Intellect cronies have proven their abilities to deliver national infrastructures many times – just look at the NHS National Programme! What could possibly go wrong with a government internet cloud?
The G-Cloud is the crowning jewel of the wrong-thinking enshrined in this report, telling you just how bad an idea it is.
Government should stop trying to build huge infrastructures and work out how to use what already exists both inside and outside government.
Biggest problem in cloud computing?
Data security.
And this government’s record on data security?
Please don’t believe the hype. Even the most fervent private sector ‘cloud’ adoptees keep their data in the most secure environments money can buy.
Martin – I’m sorry but you appear not to understand what cloud computing is. Cloud computing is not amazon Ec2 or Google (which – is where David Cameron is rumoured to want to store our health records btw) – it is a model of computing in which resources are virtualised.
This section isn’t proposing to throw all your records onto Google – it is talking about creating an internal private cloud for government use.
Comments such as yours seem to suggest to me that this “crowd sourcing” attempt by the Cons is INNAPROPRIATE and UNNACEPTABLE – If the Cons want to formulate an alternative policy they ought to listen to people with real expertise, and not every T, D and H who can schlepp onto a website and rant like and angry tramp.
Why the idea of G-cloud could be a good idea, if built on solid foundations an slowly built over time it could work an it seem like we are moving in that direction. It also could in the long term make it easier to implement upgrade.
However the biggest problems the governments has is planning an implementation. I do not see how they could even complete there planning in the time frame they are saying yet along developed a prototype by 2010, it just sound like a other poorly planned project to me, going along with a new buzzword they heard in the media.
But at least both the conservatives an Labour agrees cloud is a good idea.
As someone with expertise in computer security (I’m a former CERT contact point for one) I’d say that even a Government only cloud has huge security risks – because stuff moves around the cloud so you get the risk of stuff leaking between a compromised cloud user and others. Right now a compromised social services computer is not good but it can’t also see into the treasury.
That can be managed but only by good system level design.
A second problem is that a cloud implies all data can be accessed from all physical systems (so they can do their job anywhere). That’s fine for things like google data but a whole different kettle of fish when you are dealing with medical data or military strategy.
That is a very very hard problem to attack and almost impossible to fit to the goals of minimal interfaces and access. You can still build mini-clouds which are attached to the data they process and which are internally dealing with only one type of data securely.
“I’m sorry, you can’t pay your council tax. The cloud is down.”
“But you’re going to send me a nasty letter, right?”
“Of course. Because the cloud has no emotion.”
Tut, dull – delete.
There is absolutely no point in using buzzwords like “Cloud”, which implies data sharing, if on the ground there are no agreements about how data will be shared. The strategy will fail immediately. The best example of this is NHS and Social Care. Huge problems and no commonality of approach in shared information. This needs fixing before any technological layer can work.
Underlying this there is a grain of good sense – govt bodies tend to want to have their own individual flavour of things which could be standard, and the taxpayer ends up funding this diversity but getting no obvious return from it. Govt is slowly getting better at standardisation, and this could help it to continue on this path, and it might help some govt services improve in some ways.. But surely it implies establishing some more central control and delivery mechanisms?
And that raises the point of how big a set of interconnecting projects could this strategy spawn? Might it be too big a set to implement successfully with sensible timescales and costs? And is it really likely to save money?
Not feeling convinced yet.
Was someone really paid to write this rubbish?
The Wikipedia entry on Cloud Computing is better researched, better written and doesn’t insist on pushing a particular – extremely expensive – vision.
Given the ignorance demonstrated by the author on information architecture and security, this is just going to be another money pit.
“Work on developing the G-Cloud strategy will be initiated in the autumn 2009″
Does this mean the work has already started (in secret)?
Government uses a lot of string. And clearly it has its own distinct requirements, as well as massive purchasing power.
Our vision is that government develop and implement a G-String enabling framework that enables public bodies to source string from a secure, resilient, flexible and cost-effective service based environment. It can use the G-String to conceal the poverty of its ambitions.
A private cloud for government? In my day we called that a server.
Gov has been using private-cloud equivalents for many years, it is has been a cornerstone of IBM’s business since the 60’s. The information assurance lessons learnt are equally applicable to x86 virtualisation (which is what the current Cloud buzz is all about).
On the one hand the author admits that the Cloud Computing model is in its infancy but other hand they say that they are relying on it for most of the anticipated savings from this strategy. Betting your business on unproven technology concepts is rarely a good idea.
As mentioned previously, there seems little understanding of the way services have to be redesigned to fit into this model, and for the bulk of the government’s existing IT estate this will probably not be worthwhile, except as the very long term renewal cycle allows.
One recurring theme in large-scale IT is grandiose schemes that are going to be transformational but instead turn out to be incremental instead. It would be much better to forget the hype, recognise the actual costs and pace of transition and work on that basis instead of the ‘with one bound they were free’ approach.
“One recurring theme in large-scale IT is grandiose schemes that are going to be transformational but instead turn out to be incremental instead. It would be much better to forget the hype, recognise the actual costs and pace of transition and work on that basis instead of the ‘with one bound they were free’ approach.”
Most sensible comment I have read on here.
Re:William says:
>>“Work on developing the G-Cloud strategy will be initiated in the autumn 2009″
Does this mean the work has already started (in secret)?<<
That's Autumn 2009 on the Home Office DB system, it's actually Spring 2010 on 90% of the government departmental systems and Summer 2011 on the rest (except the F&CO, where it's June 1st 1963).
Unfortunately, that's about as compatible as the various government department database systems currently are.
Kudos to Steve Horgan for making most sense here.
We need to rise ‘above the cloud’ – Web 2.0 is changing business models dramatically, and the Cloud concept is the response to this change. Google is leading the way and as the ‘clash of the titans’ (google vs Ms) gets underway the public sector like everyone else has to place its bets…if we end up in a Google web-based, world then it will not be ‘if’ a cloud but when and all the issues to do with security, ownership and portability of data will just have to be dealt with….
How about going the other way, small-scale, well-bounded, federated systems which can interoperate. This is bio-mimetic because it is cellular and does not produce huge unaccountable, unmanageable monoliths (we see no details of the G-cloud, but civil service culture + the usual suspects will ensure that it is large and complex and unmanageable). The small well-bounded systems have good ownership and can fail (remember our banks? there used to be many, smaller ones) from time to time.
The common parts are standards, languages, software components and applications which are reduced incrementally, again so that ‘bad’ calls can be backed out.
I also agree with other commentators about the blurry nature of the G-Cloud (a private EC3, a huge REST accessible hash store etc. etc.) until will know this, we can’t call the security and point of failure concerns, But I am already concerned by the ‘handing off’ of management and ownership. Too much of that already.
I completely agree, Hugh — modularity is the key, and with systems built on open standards and FLOSS interoperability between modules would be assured.