NATS climbs into the cloud to fight legacy software snafus The National Air Traffic Service is shifting its software onto a bespoke cloud infrastructure, which it hopes will help reduce delays the next time it's hit by a systems outage similar to the one it suffered last year. The body is spending tens of millions of pounds on its "aviation cloud", it says will improve "resilience and …
Well they're quite clear that it's on premise, so under their direct control.
I totally agree with you however, WTF is meant by cloud in this context. Smacks to me of a PR bod using buzz words to garner some free column inches. There's probably a NATS IT bod squirming somewhere.
In fact what is probably happening is that NATS will replace some old onsite infrastructure with some new onsite infrastructure, probably deploying machine virtualisation so that they can scale up capacity super fast in the event of a spike in load. This would make sense given the last failure resulted from overloaded server hardware due to a spike in load caused by an unforeseen use case of the NATS software.
I totally agree with you however, WTF is meant by cloud in this context. Smacks to me of a PR bod using buzz words to garner some free column inches. There's probably a NATS IT bod squirming somewhere.
Interior: Accenture. All the customers are hipsters. NATS lands - downwards (on fly-by-wire).
NATS: Morning.
PR bod: Morning.
NATS: What have you got, then?
PR bod: Well there's Win32 and POSIX; SQL, XML and POSIX; Java and cloud; Java, POSIX and cloud; Java, POSIX, XML and cloud; cloud, POSIX, XML and cloud; cloud, Java, cloud, cloud, POSIX and cloud; cloud, cloud, cloud, Java and cloud; cloud, cloud, cloud, cloud, cloud, cloud, SGML, cloud, cloud, cloud and cloud; or a custom-built Oracle backend running on several server instances accessed by an SQL Forms user interface and cloud.
NATS: Have you got anything without cloud in it?
PR bod: Well, there's cloud, Win32, XML and cloud. That's not got MUCH cloud in it.
NATS: I don't want ANY cloud.
Etc... etc...
They're obviously undertaking an infrastructure refresh and delivering that back in to the business as a managed private cloud service. It provides a different consumption model for the various elements of the business to adopt.
In the 90s and 00s - IT departments were tin buyers. Now they're turning in to service providers and I guess NATS are the same.
"Love to know what they mean by 'cloud'. Critical national infrastructure lives where and under whose control?"
Having a (private) cloud doesn't necessarily mean that your IT is under the control of a third party. Likewise, even if you have a non-cloudy in-house IT system, it's likely that large parts of its maintenance will still be outsourced to other companies.
I know for a fact that as of about three years ago large parts of NAT's IT Infrastructure were being looked after by a large American company.
"how does shifting to [on-premises private cloud] allow them to junk software in a way that they presumably couldn't do by changing their own hardware?"
2 days later... a very simple very relevant question remains unanswered.
"If I have misunderstood the point of the article then apologies etc"
I think you've understood more than many others. But tread carefully, understanding such things may not always be advantageous to your job continuation plan.
Into which we will shovel the reprogrammed "legacy" code converted from the old legacy (mainframes anyone?) languages running on the newest versions of LINUX, and which will be tested to work out all the problems caused by lack of documentation of that legacy. Been done dozens of times before (really) at about 1+ manhour per instruction. More if it is real-time.
Hire one of the cloud providers to monitor that several hundred boxes running the infrastructure code they understand, and a few greybeards to look-see the macro operations..
Yes, the use of the phrase 'cloud' seems to have been abused here, as said above: cloud is just "someone else's server". This is clearly a virtualisation strategy, but the word virtualisation is so 5 years ago. It is a shame that NATS IT bods have to worry so many of us by using the word cloud when they mean virtualisation.
Heathrow already flies planes over huge densely populated areas of London and thus damages the health and lives of more people than any other airport in Europe. A third runway would increase this number of suffering people to a million+. Pollution levels at Heathrow already breach legal limits. They feel paying UK taxes is a burden and avoid doing so to maximise the profits they can send overseas to their foreign owners.
How does putting a system into a cloud prevent outages? The snafu that caused havoc was a software bug, It does not matter if the system is running on real hardware, a VM or a cloud (a VM wrapped in hype"), the outage would still have happened
What at load of rubbish. Someone can now put a tick in a box and say that they are using "cloudy stuff"
The "This will stop software bugs" BS only makes sense in the context of either
a) A really deep code review of all their code, sweeping for any known bug patterns, and then re-sweeping if any new bug patterns are found
b)Clean room re-write. Blank sheet driven solely by the requirements documents. No peeking at the current code.
Funny thing. When companies realize how many person hrs have been clocked up writing this thing, they suddenly decide it's not that important after all.
This is a critical service. It hardly ever fails. So, evidently, they aren't using any windows - it'd be criminally irresponsible to use windows for anything that is potentially life-threatening.
In fact, mainly, ATC has used APL, which is why they've had mainframes about.
Yes, it sounds like another name for virtualisation. Virtualisation is also a poor solution for collections of complex systems that are all working on the same demand patterns. All it does is conceal the specific capacity problems with the long term effect of having worse capacity-related outages at some time in the future.
A more sensible long term solution would be to, where possible, move more ATC decisions into the aircraft software (Ada, so reliable).
Who the "£$%^ are they listening who are smoking G*d knows what?
NATS has proven itself as expensive and flaky with shiny new things as The Met.
I can see this decision will end in hours of hand wringling whilst consultants make gazillions!
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
