San Francisco Muni's Rail System Will Spend $212 Million To Upgrade From Floppy Disks (govtech.com)
- Reference: 0175305475
- News link: https://hardware.slashdot.org/story/24/10/22/2349258/san-francisco-munis-rail-system-will-spend-212-million-to-upgrade-from-floppy-disks
- Source link: https://www.govtech.com/transportation/s-f-muni-will-spend-212m-to-move-train-control-off-disk
> The software that runs the system is stored on floppy disks that are loaded each morning and an outdated type of communication using wire loops that are easily disrupted. It was expected to last for 20 to 25 years, according to Muni officials. It moves data more slowly than a wireless modem, they said. By late 2027 and into 2028, a new communications-based system, which employs Wi-Fi and cell signals to precisely track the locations of trains, will be installed by Hitachi, which will provide support services for 20 years under the agreement.
>
> While the current train control system operates only on the Market Street subway and Central Subway, the new system will control Metro light rail trains on the system's surface lines as well. The Hitachi system is said to be five generations ahead of the current system, said Muni Director of Transit Julie Kirschbaum, who described it as the best train control system on the market.
[1] https://www.sfchronicle.com/sf/article/muni-board-approves-new-train-control-system-19846976.php
[2] https://www.govtech.com/transportation/s-f-muni-will-spend-212m-to-move-train-control-off-disk
I remember the original installation (Score:2)
After a few times needing to walk through a portion of the tunnels as the system failed, they had finally upgraded the system to modern communications that would allow for driverless trains... just a "safety driver." Interesting that the solution was neither modular or easily upgraded as the system grew over the next few decades.
Remember! (Score:2)
Don't copy that floppy!
Wire loop? (Score:2)
Do they mean 4-20 ma systems? Modbus? Something really old?
Isn't there something like this for their system?
[1]https://www.bigmessowires.com/... [bigmessowires.com]
Granted it probably is time to rip it all out and start over. I helped do something similar at a chemical plant, but we had a 10 day maintenance shutdown to do it.
[1] https://www.bigmessowires.com/floppy-emu/
Re: (Score:2)
I believe they're speaking of [1]induction detectors [wikipedia.org] to detect when the train passes.
[1] https://en.wikipedia.org/wiki/Induction_loop
Re: Wire loop? (Score:1)
Induction loops are old school - they are moving to CBTC (as they should be if they're spending $212m on it!)
Start over - can't (Score:2)
They don't know what the current system does. I've worked on a number of these legacy systems and they have had crap slowly added to them over time and all the documentation is now lost, out of date or for a completely different system. It is unlikely that you could find a quarter of the source code of what is running. And the source code you would find would be full of #ifdefs that were likely set as compile options but no one knows which options the current image has. You would need a team of competen
Re: (Score:2)
" I've worked on a number of these legacy systems and they have had crap slowly added to them over time and all the documentation is now lost, out of date or for a completely different system."
I've been there, LOL. DEC Alpha running sequences. It was still going in 2013 when they shut the whole unit down. Then there was 1985's finest Provox system feeding it data.
Re: (Score:2)
the voice of experience
Re: (Score:2)
TFA says the system is from 1998 and already "the writing was on the wall" that floppy disk technology was in it's sunset phase. But this "going around inserting floppies every morning" really makes no sense. Flash memory was expensive back then but hardly a drop in the bucket when you are spending millions on a system like this. Hard drives were an option too (spin up, load the system software, spin down until next morning). Unless there is some details missing here?
Huh? (Score:2)
"The software that runs the system is stored on floppy disks that are loaded each morning" because desktop hard drives or network booting did not exist? How old is their equipment? :-|
This will fail (Score:2)
My guess is that this will fail because there is too much money involved. Just an itutition after having seen too much IT disasters.
Let's revisit this prediction in 10 years.
Re: (Score:2)
The reason floppies are still used in nuclear weapons silos and trains and Boeing planes is their simplicity and robustness to degradation. Making it more complex does not itself make it better just be cause its newer.
I don't think even 10 years is needed, I expect it will be hacked or have a technical meltdown within 5. We shall see.
They installed C64 REU's and CMD HDD's (Score:2)
Isn't progress wonderful? ;-)
Interim solution (Score:2)
Hit the retrocomputing community: I bet they could easily keep the current system as-is for the time being but upgrade the floppy drives to flash memory or even make them look like physical drives to the original system but pull files on the network on the other, with some el-cheapo Raspberry Pi board or something. I'm always amazed to see what these guys come up with to keep truly ancient computers going and relevant today.
Focusing on floppy disks is missing the point (Score:1)
Personally I think that the fact that floppy disks are still being used receives unnecessary focus (per the SFMTA's own words, this is 1980s tech installed in 1998, what do you expect). The more salient point is that (per [1]https://www.sfmta.com/projects... [sfmta.com]) they are upgrading the existing train control system from the older ATCS (subway only) and manual control (everywhere else, and when ATCS comms fails) to CBTC - this allows the control center to know much more accurately where trains are at any given poi
[1] https://www.sfmta.com/projects/train-control-upgrade-project
$212 million (Score:2)
Are they that fucking stupid? How can it cost $212 million? Do these morons not have anyone that knows how computers work?
Re: $212 million (Score:2)
For real. Slap some air tags on the carriage and call it a day. Your welcome.
Being facetious of course but one could think of dozens of ways to track trains and use that info to route, re-route, prevent accidents, etc for a mere fraction of 212 million.
And yeah, going back to the original vendor after 25 yearsâ¦that Hitachi sales guy has retired on this bokndoggle
Re: (Score:2)
no. no one knows how computers work. its a collective effort of thousand of people around the world keeping a complex ever evolving stack of jenga tower bricks from toppling over.
you could probably kill 10 people and setback the entire industry decades.
Re: (Score:2)
Heathen! Minidiscs are the one true upgrade for floppies.
Re: (Score:2)
Just think - they could hold ~80 boot images on a single LS-120 and only have a single point of failure.
Re: (Score:2)
oh, man, I had such high hopes for that format. Bought 2 drives and at least 5 discs way back.