It’s a long march towards realising G-Cloud and releasing consistent data across government, says Bill McCluggage – who until recently helped lead the charge on cloud. He also spoke about how government would manage the release of Open Data from its data centres and on the public cloud. The former deputy government CIO told The …
"the legacy stuff we work on goes out as well"
True, but for bigger and better screw-ups, outsourcing/offshoring to the cheapest possible operator is the way to go. We've had outages on ATM networks (and banking systems in general) in the 50+ years we've been running them, but can anyone remember one that went on for days (weeks if you're unlucky enough to be with Ulster Bank)? Then RBS sack all their experienced (expensive) staff and within months the roof falls in. Anyone else detect a pattern here (not if you're a bank executive, obviously)?
The latest 'cloud' (whatever that means) 'will fix all your IT problems' (where have I heard that line before) is just the latest example of short-term cost-reduction leading to (slightly) longer term disaster.
If this is his level of understanding of government after 7 years, then we are well shot of him. For central government, there is only one Home Office, or DWP/HMRC. They have pretty unique requirements, that are not the same as any other organisation, so they have bespoke systems. Leveraging in COTS products creates the sort of nightmares that the CSA faced when using a system from Florida that couldnt handle such arcane concepts as VAT. Yes, where a service is generic (postcode checking, input validation etc) then a cloud service is usable, but not for most caseworkers and policy makers/knowledge workers, who will really love that application latency as their screen redraws as data comes back from India, Iceland and Indiana, rather than the empty datacentre in the basement.
Square pegs and round holes
Just denying that different departments have different requirements doesn't make it true. The guy clearly knows nothing about government - it doesn't get to pick its market segment or target audience.
The guy is just another clueless beancounter if he's stuck in the foothills of data modelling and networking - there's lots of history for both (even going back to the days before shiny boxes) and the British Government wrote large chunks of it back in the days when it employed competent people. If he can't imagine a cascaded data model which allows for local descriptions then it's no surprise that he's struggling with a bit of network theory.
To bespoke or not to bespoke?
McCluggage says: "People come in and say: ‘I need to bespoke this'."
Do real people (as opposed to empty suits) actually speak like this? Bespoke is now a verb?
Enough gratuitous verbing, I say!
I saw the G-cloud stand at a civil service exhibition recently. It consisted of one screen showing a rolling presentation and a couple of civil servants standing around looking bored. The presentation was just words popping up on screen telling you how great the G-cloud is, but the timing meant you didn't have time to read half of them and on some slides, the words didn't even fit on the screen.
It's no surprise that take-up is slow when the sales pitch looks so amateurish.
What does he mean by bespoke? The addition of another table in the db structure to hold unique data for one area, with the requisite security controls to only allow them to access or a unique input/output screen for an area pulling data from various points in the db??
There also seems to be a little muddle in what he's saying - is this to be a db for creating stats or is it going to be used for data entry?? The two can be done from the same source, but it's horribly wasteful and inefficient.
Having 2 data repositories would also make releasing data much easier - an internal db with all information on it and one with external links with only statistically relevant data (so no names and no addresses below postcode). At which point the data isn't personal and can be released at any point.
let's hope that the g-cloud isn't shared services on steroids
"Shared service centres have failed to deliver the savings they should have. They cost £1.4bn to set up, £500m more than expected, and in some cases have actually cost the taxpayer more than they have saved. I welcome the Cabinet Office's ambitious new strategy for improving shared services. But unless it learns from the past it will end up making the same mistakes again."
Great article and I agree that making data accessible is the first step towards opening data. For example, we recently spoke with a CIO of a city who partnered with local developers to create a mobile app to help community emergency responders support local officials in case of a natural disaster. It would have taken years for the city to develop the app, but by opening data and working with local developers, it was completed in months. We also have seen apps for reporting dangerous street conditions, and many others including education, security and health.
Regardless if its in the cloud or on-premise, improving data sharing does not require making it all available. Start with what you already have open and then make it easily searchable, sociable, and sharable.
Commercial off-the-shelf software can support organisations in their off-the-shelf processes. However, when you you do something unique, you will need bespoke software to support it. And of all the organisations in the UK, central government departments are some of the most likely to have unique processes, and so are the most unlikely to find off-the-shelf software to support them, because no software house will develop an application for a one-off sale unless they have been commissioned to develop it.
Central government department will find three kinds of proposition in the G-Cloud. First they will be able to find off-the-shelf applications for their own support functions - HR, document management, and so on. Second, they will be able to find off-the-shelf components which can be included in new assemblies to support their unique processes - middleware, databases and so on. Finally, they will find any number of off-the-shelf applications whose vendors claim will be able to support the department's requirements 'though configuration'. Having forked over (our) cash for these they will discover that the configuration is done by writing a whole bunch of new code, or through a special application-unique language which they will never be able to hire for.
Central government department are bespoke. Why would you expect to be able to support their bespoke functions with off-the shelf-software?
Christopher Pontac (Solutions Architect)