NHS tears out its Oracle Spine in favour of open source The UK government's quest to get public services to use more open source technologies seems to be taking hold, judging by the revamp of the NHS's very large Spine service. The upgrade from Spine to Spine2 will see the NHS shift the core of its main secure patient database and messaging platform from Oracle onto a bevy of open …
> Spine, which was one of the two successful components
So they've identified one of the successes in their ill-fated adventure and have decided to mess with it. I suppose that will, at least, bring it into line with all the other badly designed, poorly managed and hopelessly implemented projects.
At least by using free (OK Open Source: not the same, blah! blah ....) software, the cost of this failure will be a lot less.
[postscript. to increase the flexibility of the system ... one of the largest backbone systems for NHS IT is this really an attribute you want in a spine - or are we talking snakes? ]
This has the trimming of a well thought out project. Not only using open source systems, the proposed
1) Takes into account the cost off the software
2) Uses a tried and tested infrastructure that can be support by a vast number of Sysadmins, rather that overpaid consultants
3) Be easily integrated with by 3rd parties.
But for some reason, I very much doubt it will stay that way, especially after it has been kicked through the Bureaucracy of 10-15 differents committees, just so they can decide on the background color!
Agree it sounds great. But the NHS open source promises are like a very obese person swearing they will lose weight next week. We have heard it before.
When big business gets a whiff of this they will smash it to smithereens with their lobbying wrecking-ball. Sorry for the cynicism.
This has the trimming of a well thought out project.
You can't tell without having seen the requirements and the dossier trying to fulfill them.
In the meanwhile, as I am full of negativity, This Image applies.
Some of the systems were built with Oracle and are now approaching or have approached the limits of what those versions of Oracle can store.
It's a very typical tactic to say you're leaving to get a better deal from the current supplier. But in this case they just want greater control over the architecture, open source is brilliant in many ways.
I've been working on a project which was going to be using a proprietary GIS server that would have required a lot of really horrible hacks and a rather lousy architecture. I've done something better using lots of open source technology that does exactly what I want, with a nice API (almost too easy) and loads of online forums and examples to help when you hit a snag. A rather neat GIS solution in two weeks.
Well, if he's talking about the mapserver software, http://mapserver.org/, it's been around longer than proprietary products Like ESRI's ArcIMS or Arcserver , is well documented, and is commercially supported by several companies worldwide. He could also be talking about Geoserver, http://geoserver.org/display/GEOS/Welcome, or Deegree, http://www.deegree.org/, openlayers, http://www.osgeo.org/openlayers, or several of the other projects listed at the osgeo site, http://www.osgeo.org/.
It sounds like you missed the International FOSS4G conference that was in Nottingham England last month. Videos of the presentations from the conference are online at http://www.fosslc.org/drupal/videos. Watch a few and judge for yourself.
The crucial word is "directly". You can't say the government hasn't learned from it's past mistakes. This time the cash will be channeled through all sorts of intermediaries, shell companies and sub-contracts. That way it will be impossible for anyone to say for sure how much was wasted spent.
So this was one of two successful components, but they decide to rip it out to replace it with a new product on a new technology stack, implemented by a relatively unknown company. And it's going to be live in 2014. Hmm.
It's not a bad idea to try some new guys and new technology - particularly if the price is right, but why not give it a go on one of the failed projects instead of one of the ones that works?
The world is full of naysayers isn't it?
When you are presented with a major screw up, you find the good things and build on them, you do _not_ throw good money after bad.
This appears to be doing that.
FWIW, Basho (who make Riak), seem to be good at what they do, and so they'd be able to get this right as far as the infrastructure goes. The application side (tornado/ python and the JS web front end) leave more questions open, but the tech stack as said here is certainly high performance and very rapid to build services in.
BJSS is quite well known in the banking field for developing high performance trading systems, so they are certainly the correct type of company to build a large scale heavily loaded transactional system.
Or would you rather a better known company, like Capita say?
I really, really hope that this goes well.
What the plan appears to be is to replace the Spine with a new version that can do a lot more, and to find a more efficient way of doing it in the process. They're building on a principle that has worked, even if it's not being done by bodging things onto the current code.
If it comes off, and vaguely on-time and on-budget, maybe some other government departments will start to question the way they go about things too.
Most any government IT project failures are not due to the failures of the chosen technologies. They are due to really bad project management. The way a project is managed is far more important than the tech that is being deployed.
Almost all successful projects are down to a few key figures who knew how to build an actual requirements set on top of whatever garbage requirements the agency developed without derailing the project or building ill will within the user groups. Good and effective project managers who can get through all the bullshit involved in government IT projects are very rare.
Seeing as how having actually useful managers in house is no longer popular, you're basically getting a random person for each project. There's no way to compare a future project with the successes of a previous project because every single variable is different.
About time this was done. I was a mere coder and therefore saw firsthand the prescott administere disaster happening.
They should have done it in stages one by one the first time. With the spine being developed and then adding the extra services to it.
Say what you like about oracle, at least they did not use SQL server and Biztalk that Accenture used which was an absolute disaster from start to finish.
The Spine may have been a success, but how much did it cost to build the kind of transactional ssytem used by literally dozens of trading houses around the world!
Indeed. Using an array of technologies which are battle hardened, and excel in their own areas for their best capabilities seems like a great idea to me.
Rather than one monolithic system managed by super expensive consultants, and with only proprietary interfaces to integrate with, as is the norm for gov IT projects.
Absolutely, instead of one bunch of super-expensive consultants, you just need ten lots of somewhat-expensive consultants, who will struggle to talk to each other and agree on common direction, all blame each other when something goes wrong, and you'll have ten companies to attempt to sue when it all goes wrong.
I really don't understand why people change things which aren't broken. Cheaper, it may (seem to) be, but the old adage 'you get what you pay for' rings true all too often. Good luck to them anyway.
You definitely need a Project Dominatrix to herd that kind of bag of cats. And quite possibly people willing to do the Death March for the Good Cause. It might work, but especially in the medial sector you may suddenly get all kind of interference from the ministerial level or the medical profession. Spine will be needed, indeed.
Agreed. I can tell you that there was no small amount of interference, sometimes for the most bizarre reasons! The trick here is to rebuild 'as-is' functionality on 'better' infrastructure, using an in-house dev-ops team as much as is possible, regression testing as you go - and take it from there. Spine as baseline for Spine2. Remember the size (overall spend) of the overall Spine contract was absolutely massive; I'm confident that the financial savings this initiative alone - never mind the qualitative improvements - will be sufficient to pay for a new hospital! That's some premium for lessons to have been learned... Let's not ignore them now. I'm hoping that this project becomes the poster boy for modern government IT.
As the years go by, and the companies providing the service/s enlist MPs' as "executive arse-wipers", you will realise that it will be, along with others, the poster-boy for another way to bleed the taxpayer dry to enrich the rich further.
With inbuilt back/front/side/top/bottom doors for NSA/GCHQ, and loads of other multi-letter combinations to extract data and sell it.
It does seem a rather complex and disjointed collection of stuff to use.
And, is Python (or indeed any interpreted language) really a good choice for something of this size? I'll admit, I've never used Python so I don't know what it is capable of, but unless you can compile it down to native code (and maybe you can, I dunno), it doesn't seem right.
Its describing all the bits of an entire stack explicitly rather than just saying 'we used oracle'.
The original oracle solution will have all of these bits too, just wrapped in proprietary boxes, or possibly as hardware (eg, a hardware load balancer rather than HA Proxy)
On python, the vast majority of time spent in this style of applications is in IO, normally with a database or messaging system. The application language is very rarely the cause of a slow down, as its not doing anything particularly algorithmic.
Witness the rise of systems like Node.js that solve this. Javascript is really quite slow (test it, please!) compared to other options, but the app framework is built to handle IO more intelligently, allowing far larger scale systems than the alternative systems with a more traditional threading model.
This post has been deleted by its author
It is very difficult to patent software in the UK or most of the world for that matter which is why pretty much all software patent cases are in the US (District of East Texas in the main). More importantly such cases can only apply to breaches pertaining to use of the breaching software in the US not elsewhere in the world.
Best any troll can hope for in the UK is a copyright claim.
(IANAL)
You have to admit that a socialist government making Bill Gates a saint or lord or whatever was obviously a stab in the back for society. One has to wonder what a Tory government is doing meddling with the finest thing ever to come out of socialism.
Kicking at the scumbags at the BumBoyClub perhaps?
But they have all gone to golden parachute land.
Can't be meaning to tackle those shitty stations at the Job Centre's next can they?
Lots of American and even a Russian company, but not any British software developers mentioned ( I'm not counting implementation consultancies).
Furthermore, non of what is mentioned has anything to do with Health.
Coming up with fairly arbitrary database and infrastructure products excludes the use of many of the available products, that are known to be effective, without huge rewriting and adapting.
The NHS claims unique requirements based largely upon scale, well having looked into some of their requirements, most of their problems actually appear to have arisen from having specifications drawn up by completely clueless individuals i.e. the ones who could be spared from their main job to spent an eternity agonising over prescription systems, for example, that don't even take into account the patient's existing medication, but do try to make the prescription process comply to some overarching medical order best practice knowledge base..........
Have a look at the Riak website, and I quote
"The National Health Service is the publicly funded healthcare system for the United Kingdom. Riak is being used as the IT backbone for this system, to help drive efficiency quality of care improvements. Riak is replacing a legacy system, which allows them to drastically cut costs while improving the performance and reliability of the system."
That was quick! You might even think optimistic.
It could cut through the nhsmail is more secure nonsense and be an enabler for encryption, digital signing and could be used nationally as an NHS P2P identification factor.
Writing as someone currently battling with Vodafone (formerly Careless and Worthless) - they run the nhsmail heldesk - and trying to get them to explain to me why they delete my outbound email and then don't even do me the courtesy of notifying me that they've done it .... I have limited faith that the national infrastructure will ever really work.
Shocking really. I wonder if this means they'll be improving the documentation on the opensource projects they take from .....
The rumour mill would be wrong on this. There are plenty of companies which provide service contracts for the technologies which have been selected for this project & according to presentations that I have seen the agreements are already in place.
The cost savings on these ongoing service agreements & the savings in annual licensing costs mean that the project will pay for itself in a matter of weeks.
And where will they find the dozens (if not hundreds) of neck beard admins required to maintain this infrastructure.? A friend of mine has spent 18 months looking to hire a MySQL admin with a salary well north of $160k. I agree that this approach to technology is exciting and high performance, but trading houses can hire and keep very well paid admins. I'm guessing the NHS will have trouble competing in that very limited market
To steal and snoop them.
Yay for that.
And Choose & Book" is a "success"?
Yes I like the open source tools shift.
No I didn't want the project in the first place, and I'm very suspicious that none of the "The Usual Suspects (TM)" have been. mentioned. They will be sniffing round like a sailor on shore leave in the red light district.
I love this initiative. It was begun in 2012/2013, and last I heard it was going spectacularly well. Yes, it's built around diverse technologies but in some ways this is an indicator of the maturity of the team selecting them and building on them. BJSS is a small player perhaps, but a highly reputable one, and what is more it has been brought in for the right reason, which is to supplement a small but skilled internal agile dev/devops team. Yay
Admittedly we (the NHS) are behind the times, but when there's a whole flotilla of various medical systems, with varying requirements that have to be hosted we have to go for the base level browser for all the applications. In the trust I work for we have been rolling out IE7 for the last year and are now progressing to IE8; the ending of XP support next year will force a lot of NHS trusts to go to Windows 7 (with the added impact of newer browsers) and that is something us support staff would welcome.
I'm not saying the NHS is at the forefront of technology but we are taking those progressive steps whilst maintaining an infrastructure that currently works, albeit not at the head of the queue singing and dancing but plodding along doing the work day in day out.
I see a few people complaining about silly names in IT, without realising that IT in general is peppered with silliness.
Really, iPad? Am I supposed to wear the thing, stick a parrot on one shoulder and shout "ARRR, JIM LAD" at random people?
Hey, what should we call this phone that's going to be used by a gazillion serious business-types and their workforce? Oh, I know: Blackberry. That's real sensible. Let's give all the workers blackberries. Yep.
Windows? You're going to install windows on my computer? Surely that's the sort of thing that hardcore gamers do to their PCs so they can show off the triple-SLI water cooled GPU rig that cost them two month's wages?
While we're at it, what kind of a name is COBOL? Was someone reading really bad Space Opera when they thought that one up?
Smallest piece of information you can store in a computer? Let's call it a bit. Oh, you want a name for a collection of bits? Well, a byte, obviously. Half a byte? A nybble, of course.
Apple Macintosh? Quark Xpress? Acorn?
Yes, a silly name is quite enough to have an otherwise very useful product, company or service laughed out of the city.
... containing your medical records ... oh alright, I'll tell you.
None.
The Riak documentation on security says (and I only slightly paraphrase) " "write it yourself". Which is the same as not having any really. But as far as I know they didnt implement any anyway, so I suppose the product fitted their requirements perfectly.
Management of this rats nest will be interesting as well.
FWIW, according to what's been published so far, cost of this "free" software is £20M to implement >>so far<<. They are keeping the cost down by not documenting anything though, and that helps get new code pushed out faster. Well, thats what they said at the conference where they discussed this a few months back anyway.
Reducing project costs by not documenting them well is in no way exclusive to OSS. Bad and/or arrogant practices are everywhere.
As to your costs statement; at scale there is little cost savings over closed source implementations. The licensing fees are obviously lower, but the increased HR costs and the longer hiring timeline pretty well negate those savings.
Going OSS shouldn't be about saving money, that isn't going to happen. What does happen is that you've got better security options as well as greater latitude for extensibility. Done well OSS projects allow you to fit the software to the business and business rules. Closed source projects often require a substantial amount of change and/or workarounds to business processes.
If the project is well managed you end up with a system that is better suited to its purpose for approximately the same money as a closed source alternative.
Yes, but if this is really part of a modular system written to a standard then should it prove to be a pain in the ass to support, it can be rewritten or replaced without spending £20 billion. Getting a dozen implementations written by smallish companies will still be cheaper than getting one written by the usual suspects.
How many projects failed under the stress of total overhaul. They should build in sections. I'll bet they put the contract to end when their system/software maintenance contract is up. Way too soon. The present system contracts are probably adjusting their monthly/yearly contracts agreements for when that day happens.
Reference: http://www.pcworld.com/article/246647/10_biggest_erp_software_failures_of_2011.html
Modular building is most certainly results in the better end product, but does not provide the best options for expensive billing. Vendors don't want to built a part of something because obviously, the total costs are lower, but they also miss out on making themselves the complete system experts. Kind of building their own monopoly.
Why Open Source is automatically fêted as the best solution with little or no opposition? Surely there are as many terrible Open Source projects as there are great ones? I'm not saying there *are* or that FOSS\GNU\GPL\whatever is bad just that I'm surprised how many people seem to assume that it's the right choice because 'Oracle = Bad and FOSS = Good' without seeming to question if this is truly the case.
Oracle = Bad because of all the bad stuff they have been accused of in the past. Search El Reg forums for details.
FOSS = Good not in every case but in the case of the specific FOSS projects have been selected from the vast range available.
It's may look random but there are several comments here by people who have experience of those projects which are generally supportive. There are are a lot more by people who don't profess experience of anything and prefer to remain anonymous who are generally dismissive.
what bothers me is how are the various bits of front-end software from Emis, iSoft and others going to interface with it?
all very well introducing a new Spine, but you may be forcing a rewrite of the patient booking and recording systems as well. And for some of the older systems thats not going to happen...
Having had the misfortune to spend a few months working on the Spine for BT I'm quite surprised to see it was actually a success. It struck me as a proper crock of shit when I was thrashing around trying to implement database calls in some Perl scripts with absolutely NO documentation about the schema or indeed the rest of the design apparently existing.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
