The perils of classic hardware doing important things

[GL1zdA]

Vancouver's SkyTrain system has been upgraded at least twice in almost 30 years but at last check it was running on three IBM Industrial Computers and OS/2.

In Poland there was one train station still running an Odra in 2010 (over 30 years in service).

[japes]

While I find the PC servers run mostly fine. I even find linux usually runs fine. I just don't trust people to properly write software to run trains on the systems. I'm sure it would be fine, running some PLCs that even back end to some Windows thing that reboots every Wednesday and sometimes survives Patch Tuesday and sometimes is gone for a week and maybe the displays at the station don't know when a train will arrive for most of a morning while people try to fix it. I'm sure the PLCs will continue to do what they're told unless the contractor screwed up the code last night and left an entire decision tree out that was really important for the evening commute.

What I find frustrating now is companies, and governments too, don't employee people for long term. Or employees don't seek a career in companies for a long term? Which follows on with companies not wanting to learn how their damn stuff works. They just call the contractors to fix the control system that makes the lights work in the station because we don't have time to learn this expensive system we just bought. Or it's too expensive to train folks.

Sadly the contractors are in the same situation, the employees are only around for a year or two. It's too expensive to learn how this stuff works, so they figure it out as they go along. No one really ever learns the implemented system and when the next high level exec comes in and finds out the system is 5 years old she demands an upgrade because how can we support this legacy system.

And the cycle continues.

Or the system that was supposed to save money turns out to burn out the bulbs faster than they can replace them and doesn't turn the lights on when people are around and soon enough the expensive system is bypassed and the lights are on all the time. Who cares about saving energy and we'll let that expensive system rot.

Soon it'll be a miracle anything works at all. I just used lights as example, but imagine any modern system.

[pentium]

^ 20 million ? pffft Oracle could do it for a mere 200 mil and the system still wouldn't work !

Ain't the Private Sector grand ?


If I were Hongcouver, I'd be afraid. Very afraid. After they upgrade this system with COTS Dell boxes and new wiring hand-made by pygmies in Botswana (it's our fiduciary responsibility to save money and bring the consumers better value !), all running Windows Vista (it was on sale), the upgrade won't go six weeks between total burndowns

Well actually, in the last year they did upgrade the scheduling computers to Wintel Dell machines, so you aren't TOO far off, mind you the system will operate on a failsafe if that system goes down.
The control room has changed a lot since 1986.

Speaking of failsafes, not three days after the breakdown someone managed to flip a breaker 10km away. It wasn't a hydro utility breaker. Just a private breaker operating part of the transit system. Through convoluted wiring arrangement this cause the entire computer system at the control center to lose power but not trigger the diesel backup to start. With over 150 trains requiring constant synchronization with the central computers, it took another three hours for all of them to be manually synced into the computer before they could put the system back in automatic.
We ran flawlessly for decades and if anything, the upgrading made the system more unstable.

[hamei]

Well actually, in the last year they did upgrade the scheduling computers to Wintel Dell machines

Did a little reading .. okay, they are blaming a "computer glitch" for Failure One.

Am I correct that the system ran without any major malfunctions for approximately thirty years under OS/2 and three antique IBM "industrial" boxes ? Then they "upgraded" to Windows / Dell and within a year had a major disaster with the entire system down, no communications, people exiting the trains and walking alongside what may or may not have been 600v live tracks ?

Interesting.

http://www.theglobeandmail.com/news/bri ... e19711156/

Then ...

... not three days after the breakdown ...

Three days later an electrician on a spur track under construction, not even in service until 2016 allegedly tripped a circuit breaker which once again brought down the entire system ? So they suspended the electrician ?

Very interesting.

We ran flawlessly for decades and if anything, the upgrading made the system more unstable.

Of course. The people running things now are imbeciles.

I don't know if you've noticed, the one thing about the Inner Party is that no matter what, they always get their salaries.

This ought to be fun to watch ... from my cave in the wilderness

[Pontus]

How many breakdowns did the old system have during its first year?

[pentium]

I did a university paper on this so I know a lot about the first five years of service.
They ran the system for almost six months in every possible configuration before the official opening. Scarborough opened using the same system a few months before Vancouver did and several major bugs got weeded out there before Vancouver went into public service (phantom trains, power rails icing up, etc.). No major interruptions occurred in the following year (which was nice because it was a major component of Expo 86) aside from malfunctioning door buttons making the computer hold up trains at the platforms. Considering that three years previous the entire ICTS technology was still a prototype with no existing system beyond a 2km loop of track outside of Kingston you really got to wonder how Vancouver winged it and got away with it for so long. Most automated transit technologies were dead-ends but ICTS (now Innovia ART) systems are still being built.