Nearly half a trillion pounds in tax and other revenues lining the UK government's coffers every year are processed by decades-old IT systems - and the National Audit Office is worried. The auditors, who this week published fresh research [PDF] on public-sector tech, estimated that £480bn of government revenue relied on legacy …
Legacy is good for serious work
"The auditors, who this week published fresh research [PDF] on public-sector tech, estimated that £480bn of government revenue relied on legacy ICT in a year. "
Good news! (Definition of a "legacy system": one that works).
Re: Legacy is good for serious work
Legacy often is seen as "old and clunky", whereas cutting edge is seen as "new and shiny", in reality it means stable, predictable and reliable with (often) most of the bugs ironed out, as long as the bits that wear out are looked after, and it still does the job there's no issue - why do you think XP SP3 is still popular?
I wouldn't be surprised if several large banking systems that have trillions going through it (per day) still depend on systems that are considered "legacy", and they only go wrong when you try and "upgrade"... cough... cough... not saying anything else (where's my COBOL manual?).
Re: Legacy is good for serious work
"I wouldn't be surprised if several large banking systems that have trillions going through it (per day) still depend on systems that are considered "legacy", "
Funny you should say that....
There is a (possibly) apocryphal story of a US bank in London running some core apps written in Univac assembler.
They went IBM mainframe later.
So they ran the apps in a Univac simulator.
Around Y2K they were running the simulator on a 3090 in 360 emulation mode.
Of course that was a long time ago and they've probably scrapped it by now.
Re: Legacy is good for serious work
Running original software on (carefully-proven) emulators is the best initial step for major hardware upgrade cycles. Your example London bank has very clever IT managers.
On the other (tree stump) end of the intellectual spectrum is found the naïve approach of switching over to SAP.
So make them simpler, duh
'"The risks of legacy ICT will increase over time as the gap between the system functionality and business need widens and the complexity of the systems and software increases," the National Audit Office (NAO) noted.'
And they seriously think that moving all the insanely complex logic to completely new systems is going to help?
The obvious answer is to make the damn systems LESS complex. Which our wonderful government could easily do by simplifying and rationalizing the laws. They won't, of course, because that would mean eliminating the multifarious deliberate loopholes that help their rich buddies to get continually richer at the expense of ordinary taxpayers.
Doesn't sound that bad to me
The figures given suggest less than 0.5% of turnover is spent on running the system. IMO that sounds like a pretty good figure. And there's that old argument that if it isn't broken, we should be fixing ... err, modernising ... it until it is.
See, it is possible to have "well managed and … stable platforms" in government IT. On the other hand, these are old systems that would have been built back in the day when people cared about such things.
Legacy tech vs tech that goes in the bin before the dust has had chance to move in? I know which I would go for. Yes eventually things need replacing and it is smart to do it prior to things going wrong, but there is a very valuable lesson to be learnt relating to why a certain system is still working 20 years later yet the latest billion pound IT disaster is headed to fleabay.
One of my first IT jobs was at a county council in the mid 90's. Next to a shiny new wolfcreek and a fancy 30ft long laser printer was a bunch of old ibm gear, tape drives (somewhere around 120mb per tape iirc?? i probably dont) and toggle switches etc. Very 1970's but frankly, it worked. They had a variety of legacy systems and managements position was the system is solid, it is paid for and we have an ample supply of spare parts and trained staff. Who in their right mind would trash it for an expensive, shiny unknown? Many of their newer systems were written in cobol and running on sanely priced off the shelf servers with adequate redundancy and backups. Their plan for replacing legacy systems was basically that, an in house written, no frills replacement.
Legacy is what our IT is supposed to be
All these billions spent on IT systems and someone is worried that it is getting old ? What, is there not a janitor in charge of dusting the things off ? Is there not a maintenance contract for replacing the capacitors that blow ?
It's about time that we get out of the PC upgrade treadmill mentality. Code does not age. Make a computer that lasts a thousand years, and your program will happily chug away for the same amount of time.
Our governmental systems SHOULD be legacy. The principle of gathering taxes is quite old, I would think that adding new tax lines is not something that requires a rethink of the whole system. The principle of redistributing taxes is also quite old, but I gather that governments do have a tendancy of doing that rather willy-nilly and without great planification, so I would not be at all surprised that there is not a clearly defined procedure suitable for handing an additional billion or two to some MPs personal buddy.
But that still does not mean that we should tear out all our systems and put in new ones every time the government changes (although the contractors would simply go berserk at the idea no doubt).
What happened was...
The old systems worked. They continued to work. Their tech was, long ago, *way* outdated in many cases. People have been writing non-working replacements in the language of the month and those replacements have failed.
It *is*, in my opinion, possible to replace these old systems with robust modern systems that are *more* stable, feature rich and maintainable. However, solid bread and butter software development is not sexy enough to sell.
Most of the world's working code is essentially written in Assembler, COBOL and C. When was the last time you saw a job posting for those? You don't because the existing systems work. Meantime, there is always a call for stuff like Java and the web tools of the month because the systems keep breaking down.
I dream of a day when actual software developers are consulted about software development. I am afraid it will always remain a dream.
*All* good systems are designed to *expect* change
Perhaps most of these were written before written before most civil service IT got outsourced and the people involved expected to be maintaining them till they retired?
Interesting to contrast this with Universal Credit
£200million spaffed ... so far
Seems like only yesterday the Government Digital Service were claiming all the credit for this project. Now it's "nuffin to do with us guvner"
The C word
In the last few years I've had customer service systems-related problems with HMRC, TV Licensing, Royal Mail and E.On. They've all been outsourced to companies beginning with "C". Maybe NAO should take a look at that.
PSN Code of Connection compliance?
Given the pain that the Cabinet Office is imposing on other aspects of Gubmint, especially local authorities, to be compliant with its "zero tolerance" stance on compliance with the ever-developing Code of Connection that requires Public Sector bodies attached to the Public Services Network (PSN) infrastructure to be scrupulous about addressing new and not so new vulnerabilities on their systems <long breath> - how come central Gubmint depts. get away with retaining what sounds like very old IT, presumably containing vulnerabilities for which patches will never now be produced, to deliver what must be considered critical and sensitive services? Eh? Well?
Re: PSN Code of Connection compliance?
One of the best things about really old legacy systems is that they are not, and cannot be, connected to the Internet. In most cases there is no good reason why they should be, and any connection - even indirect - is a latent vulnerability.
- Analysis Ballmer's billion-dollar blunders: When he gambled Microsoft's money and lost
- Blighty tablet sales plunge 31 per cent in saturated market
- Q. How much did Google just spend applying political pressure in the US? A. $4.6 million
- Oracle storage analytics break Oracle storage appliances
- Daisy completes Phoenix IT gobble