Gov.uk Platform as a Service
cloud.service.gov.uk
cloud.service.gov.uk
[0] https://www.gov.uk/government/organisations/government-digit... [1] https://design-system.service.gov.uk/
Edit: Yes, I know the UK Gov's system for benefits isn't easy to understand. That's on the Gov, not GDS :)
In my experience gov.uk has been vastly superior compared to other websites for a long time. Gov.uk seems to be more consistent with everything than most company websites are.
Isn't that massively excessive? Why does the Prime Minister need to be spending his time listening to a report direct from a team implementing software services?
With a direct line to the PM, every time a mandarin tried to block things (likely in order to favour his golf buddies from this or that vendor), GDS could go "ok, shall we take it up with the PM then?" and that would be the end of it.
It's effectively the nuclear option, in terms of civil-service management, but sometimes it's the only way to get stuff done.
Which is another way of saying it conveys the reality more realistically than a documentary could ever get away with.
There seems to be a much longer history of MPs and PMs being corrupt than the mandarins.
Not that I'd attribute it to any PM, believe it was driven by the government security services (which department/flavor I could not say).
That all said, it may of been a different aspect but certainly has a longer story than told. With that, it started as an intranet project iirc.
https://gds.blog.gov.uk/story/ has some insight but that seems more to cover the more public web aspects and even they say it is "a story" not "the story".
So be fascinating to peace the "full story" and history together.
GDS was a whole catchment in the early days, that included the phone services and that was contracted to Mercury telecom at one stage, or aspects were. https://en.wikipedia.org/wiki/Mercury_Communications
If any credit should be laid, it would be upon the civil servants and associates that came up with the initiative, who sadly will probably remain nameless even if their names were known. Politicians just green light so much and claim credit, so much goes on behind the scenes of any government. With that, best think of politicians as orchestra conductors who depend upon musicians knowing how to do things, often able to do things right without eye sight of the conductors at times.
I don't really like giving them credit for anything but there it is.
> it would be upon the civil servants and associates that came up with the initiative
From what I understand, the power of this was precisely that it came from people outside the career civil service. Although this narrative came from Conservative quarters (who occasionally indulge in civil-service bashing), I find it credible. Established powers in the system tend to have established friends among deep-pocketed vendors, particularly these days that doors tend to revolve quite furiously. For all the good they did elsewhere, one undoubtedly bad legacy of the Blair-Brown years was eccessive trust on friendly vendors vs in-house.
Martha Lane Fox, bless her.
And the brilliant Tom Loosemoore, also behind https://www.theyworkforyou.com/
The GDS started well, but I don't think the GDS can continue to cruise on its earlier reputation. It too has started to fall back into the bad old days of massively wasteful projects with unclear scope, such as the Verify system, that is absorbing hundreds of millions of pounds for dubious benefit [1].
"Unfortunately, Verify is also an example of many of the failings in major programmes that we often see, including optimism bias and failure to set clear objectives." [2]
I think the idea of GDS is a good one, but evidently it needs to continue to reevaluate itself to avoid regressing into what it was originally setup to avoid.
[1] https://www.computerweekly.com/opinion/Back-to-the-past-with... [2] https://www.nao.org.uk/report/investigation-into-verify/
Actually the credit should go to the brilliant and dedicated people in GDS doing great work; though I do not know enough to pick out names. The UI/UX and implementation was first class from the get go.
As a Brit my life is made remarkably easier as a citizen and startup entrepreneur. The benefits to the UK ecomnomy are enorous but not measured. In contrast my personal experience of govt engagement when setting-up and running startups in US between 1991 and 2013 were painful. And would have been very painful without help (thank you YC).
Finally if we are going to give leadership credit it should go to Francis Maude [0]. He happened to be my local MP, so I met him quite a few times, and he was most helpful and astonishingly effective in helping us navigate two big challenges we had in facing down, shall we say, "unethical quango-business partnerships". A remarkable, thoughtful man and effective operator, who was in government over three decades [1].
[0] https://gds.blog.gov.uk/2012/10/16/gov-uk-the-start/ [1] https://en.wikipedia.org/wiki/Francis_Maude
Martha Lane Fox also deserves mention for: https://www.gov.uk/government/publications/directgov-2010-an...
They all tore up the rulebook. The rule being: Government can’t do computers. Outsource to favourite big consultancy. Twenty years were wasted on that nonsense approach.
Some of the greatest resistance comes from the big departments who are actively hostile to many of the changes required to work in this way. There’s a lot to be said for a round of golf at The Belfry with the splendid chaps from the IT consultancy your department spends £2bn a year with.
Golf days: https://parliamentlive.tv/event/index/abfe49d3-f24c-4b93-b8e...
Recently I was with a German friend and she nearly fell on her ass when I showed her that you can actually pay at a car parking machine in the middle of nowhere with Apple Pay here in the UK.
It continues from strength to strength.
However its pleasing to see a site that works well on crappy computers.
Some important UK government services are still lagging behind. DVLA, since you mention it, has a mix of okay and truly appalling services. I'm in the position of trying to renew my driving licence after it was revoked for medical reasons. It's now six months since my cardiac arrest and I now meet the medically-fit-to-drive criteria. But, I face an 84 day delay in my application being processed, as detailed on the DVLA's 'how-slow-are-we-today' page [0], which is actually labelled as a COVID-19 update. Furthermore, I have to post a paper application form to them as they haven't yet digitized the medical review service [1]. This is life-wrecking stuff for anyone who needs a driving licence for their work or similar.
I do like the gov.uk sites in general, but a lot of the back-office stuff is still disastrous. Some of this is due to strike action and COVID, but a big part of the problem is the continual failure of the various UK Govt departments to successfully manage large IT projects.
[0] https://www.gov.uk/guidance/dvla-coronavirus-covid-19-update
[1] https://www.gov.uk/reapply-driving-licence-medical-condition
To get a decent result on any IT project you have to fix the organisation behind it before you even think about replicating that organisation digitally.
Exactly. I've seen this myself back in the day when I supported the MOD's system development (simulators for training, as it happens). As a techie, I was asked to develop some interoperability middleware between disparate systems. As my career matured, my focus changed to trying to get the systems to use common interfaces protocols / standards. Then I thought - why can't they have common requirements since may of them are doing very similar things, and drive tech commonality top-down. So I started working with the respective acquisition teams to get them to adopt common conventions. Then I started to question why they had disparate acquisition teams in the first place. This turned out to be because there were multiple customer project teams. Why was that? Because the different tribes in the army (infantry, armour, etc) didn't speak to each other at that level.
I eventually realised I was seeing a sort of example of Conway's Law [0], and was finally able to develop a proper briefing for a senior decision maker who was concerned about the technical / contractual complexity at one of his training schools (which had multiple different types of simulator, each with out own through-life support programme). This I believe finally led to some coherence-focussed reorganisations of the MOD teams involved, which was perhaps the most impactful thing I achieved in a couple of decades working in this space.
Part of the problem is that they have multiple back-office legacy systems. I remember reading recently about a modernization of the personalised number plate system. This was apparently completely separate from the licence management so could be modernised quite easily. How convenient that it just happens to be a good revenue generator for DVLA, so presumably it was easier for someone to justify the development costs.
Where've you been?
Agree with your points but the things you mention have been around for almost 10 years now (driving license/passport renewal easy UX etc). I renewed my passport around 7 years ago and it was much easier to use then. It's true there's some services that tail off into the older systems (and you can tell), but the modern things gov.uk has owned have always been good since launch.
But for web developpers, that’s not all: it’s very deliberate, there’s detailed public guidelines, a scrutiny that developpers of the most popular open source project might recogonise. The overall result is that the people working there are very good. They are principled, and typically very complementary of the independent-minded polymath that many of us have worked with. They understand accessibility in detail and many dimensions (think: both rare handicaps and aging web-clients).
I’ve worked with several and I would strongly recommend considering experience at the Government Digital service (GDS) as a very good signal on a CV, akin to senior role at a big tech, or early engineer at a successful start-up.
https://digitalservices.sfgov.org https://medium.com/san-francisco-digital-services
It’s based on Cloud Foundry, a “run your own Heroku”-type project. That’s starting to show it’s age now but it’s been a great model and had major benefits in government.
Happy to answer more questions.
We've maintained the core Cloud Foundry experience, and added value around the edge with service brokers, good documentation, support, app autoscaling (using the CF-blessed autoscaler, which doesn't come "in the box"), some additional metrics, and a couple other bits.
Just the autoscaling wouldn't be enough to keep us on it, but running it on someone-else's-infrastructure with a proper support model might.
[0]: https://www.waypointproject.io/
[1]: https://github.com/swisscom/waypoint-plugin-cloudfoundry
In more "Napoleonic" countries the State would just mandate adoption and you'd never know it, but in the UK the public sector is awfully balkanized, so there is a need for public marketing.
What does this phrase mean?
Applying this to a government you will have one where you have a lot of separate agencies that are hostile, uncooperative, and don't want to follow any unified standards.
Jim Hacker would have probably replied "So you're telling me they are not playing against each other, except when they are playing against each other...?"
Are there any other cloud providers who have started by reselling another cloud providers products? Even as a reseller they ought to be able to make a decent profit by relying on volume discounts.
I doubt the UK government is interested in setting up competing infrastructure.
Having a consistent platform like this also gives them negotiating muscle because if they wanted to up sticks and move completely it is much more feasible.
Do they?
AWS's pricing may be eye watering in a few cases - looking at you egress - but one thing they definitely get right from my PoV is rarely if ever hiking prices
That means bill increases become costlier at scale as the volume discounts are not deep enough in a build vs buy on a pure cost basis (Lock-in, ease of use, reliability, talent reasons apart) at 100's of millions or billions spends governments do not to consider inhousing.
Not really, this is not something we want governments doing.
In Canada, we spent almost $1B on a simple gun registry - the equivalent of a few spreadsheets with a few million records with security overlay, and almost $1B on a Federal Payroll system that doesn't work.
IT, particularly products, is one of those areas where budgets can go over by infinity, it's systematic white collar corruption.
If AWS becomes less than ideal, there's maybe another service provider that can do it.
Even this initiative, while laudable, may not be perfectly operationally ideal.
Just like AWS actually has carved out a special government offering of their services, I believe a host of other providers should as well ...
A PaaS/SaaS customized for a given domain, like a country, that built out basic identity services and privacy features given the local regulatory environment, would be ideal.
Then the gov. can legislate, communicate, provide oversight while risking the fewest dollars doing the trickiest things.
It's not entirely implausible that things can be managed effectively, Sweden has done some amazing things in particular with process and communication. But they are small and special and take 'design' seriously literally as a culture.
For government, yes. The USAF has a common infrastructure provider called Cloud One, built on top of AWS GovCloud. You can try to go to their website at https://aws.cdl.af.mil, but you won't be able to, since it requires a CAC and mutual TLS.
Among the things it gives you on top of AWS:
- CAC authentication, so MFA isn't optional like with vanilla AWS and you don't need separate IAM for human users (at least not the "I") since every user already has an externally provisioned and managed identity
- Air Force managed firewalls, so you can't just choose to be completely open, which vanilla AWS is perfectly happy to allow
- A limited subset of AMIs and managed services that are pre-approved as being sufficiently mature and secure for DoD production use
- Only pre-approved roles and policies, i.e. you can't accidentally create a publicly accessible S3 bucket because the system won't even allow that as an option
And yes, there are economies of scale and volume discounts and what not that have allowed some DoD infrastructure and platform providers to operate without any appropriated spending at least for some period of time, as in they fund themselves entirely by charging other government programs that use them. Platform One was like that I think for at least the first 18 months it existed, but then eventually received its own direct funding.
Maybe I'm being too cynical
I'd guess that salary is on the bottom 10 percentile for enterprise architects.
Edit
Pension is a good reason too, it is my main reason :)
From what I can see, the annual leave for these roles is 30 days per year, this is nice but 25 days is pretty standard in most tech jobs in the UK, with some employers allowing you to sacrifice salary in exchange for more days (+ or - 5 days in exchange for equivalent days of salary)
I know the pension plan for civil service is particularly good, but I wonder how it compares in terms of actual pension pay with having worked for big tech and contribute part of a salary that is, as you said, probably double – plus company contributions or matching which can be quite high (5-10%). Also, the RSUs you get and yearly refreshers kinda become your pension plan for a lot of tech workers.
Life isn't all about money, but for most people money is the primary reason to choose one company over another.
Obviously anecdotal evidence regarding the above but still
- IAM Federation and brokerage service that takes arbitrary IDPs (every application is a simple document management problem at the tip of the iceberg, with a giant authN/AuthZ problem underneath. Consider authZ as a service as well)
- HSM/Vault for standard secrets management across applications
- Encrypted backup to data lake service, with cold storage option.
- vulnerability scanning and patch management (lines of business can't be trusted to do application patching)
- CI/CD pipeline service
- Jira/ADO or other "standard" project collaboration scheme
Managers don't usually know about this stuff, and they're the dealbreakers when you don't have them.
This seems to be a national issue though, akin to the rejection of ID cards.
The flip side is that if you don't want this ID it becomes more difficult (but not impossible) to access any digital service. It also relies on an integrated public population register, which some people have problems with. But it's not a physical ID that you can be asked for or are required to have, so doesn't have the ideological & political issues of requiring a universal ID card.
There are already a ton of snowflake "single platform for X department" deployments.
That said, I applaud the effort and hope it becomes the norm.
The developers behind this effort are great, and well, they use Ruby which is my language of choice at the moment.
The issue seems to be with Firefox, there's no issue in Edge. I assume Chrome would be fine as well
That's like complaining that Amazon have a monopoly on hosting Amazon services.
> A sensible approach would have been a list of certified service providers and contact points for contracting.
No let's not for example ask the Forestry Commission to build or contract their own cloud services... that's not what they know about or what they're going to be good at.