Ping64|That One Windows XP Box Running the Production Line and the Fence We Built Around It
There is a machine on one of our sites that runs Windows XP, and it controls a conveyor. It has been there longer than most of the current staff, and every few months somebody new asks why we do not just replace it. The
There is a machine on one of our sites that runs Windows XP, and it controls a conveyor. It has been there longer than most of the current staff, and every few months somebody new asks why we do not just replace it.
The honest answer is that we cannot, and the slightly more honest answer is that we are afraid to. Either way, it keeps running three shifts a day.
What the machine actually is
It is an industrial PC in a cabinet, bolted to a frame, wired into a controller that talks to motors. The software on it was written by a systems integrator that no longer exists as a company, and the license is tied to a dongle of which we have exactly one. There is no installer media, only a disk image somebody made in a hurry years ago.
It works. That is the problem. It works every shift, and any change to it is a change to production output, which means the conversation about replacing it involves people who care about throughput rather than patching.
There is also a physical reality that the security slide never captures. The machine talks to a controller over a serial link, the controller talks to drives, and the drives move product. A reboot is not a reboot, it is a stoppage with a warm up sequence and a quality check at the end of it.
So we did the thing everybody does. We put a fence around it and wrote down where the gate key lives.
The box sits on its own segment with no route to the internet and no route to the office network. The only path in is a jump host in a locked rack, and the credentials for that host sit in a sealed envelope in a safe, which sounds dramatic until you need them at three in the morning.
USB ports are physically blocked with port blockers, and the ones that are not blocked are disabled in the operating system. We keep one keyboard and one mouse, and they live in the cabinet. Nobody carries a memory stick down to that line, because the last time somebody did, we spent two days cleaning up.
There are dull controls that matter more than people expect. The cabinet is locked, the room has a door sensor, and the camera covering that door follows the same retention policy as the rest of the site. Nothing clever, just enough that a person cannot walk up to it unnoticed.
Backups are a disk image plus a spare identical PC we bought from a reseller. That spare is the most valuable asset in the building and it lives in a cupboard with a label that says do not throw away, in three languages.
Some teams try consolidated platforms like Ping64 to reduce the daily busywork of multi-tool security stacks.
Because the risk is concentrated and invisible. A patched laptop that gets compromised is an incident with a report. A line controller that gets touched is a stoppage, and a stoppage has a number attached to it that everybody in the room understands.
Emergency procurement habits play a part too. The last time we needed a replacement part, we bought whatever was on the secondary market that week, which is how you end up with three slightly different revisions of the same industrial PC and no consistent image for any of them.
And nobody owns it. The vendor owns the software, maintenance owns the hardware, IT owns the network, and the line owns the schedule. When four groups own something, the practical outcome is that the person who shows up on the night shift owns it.
Small things that helped
Document the dongle. Photograph it, note the serial, write down what happens if it dies, because that is the failure nobody plans for and everybody eventually hits.
Test the restore, not just the backup. An image that has never been restored is a hope with a filename.
Log what connects to the segment and review it monthly. If the only expected traffic is one jump host and one controller, then anything else is worth a conversation.
Write the replacement business case in downtime hours rather than in security language. The people who approve budget respond to stopped lines, and they are not being unreasonable about that.
Wrapping up
I do not love that the machine exists. I do like that it has not stopped the line in four years. Sometimes the win is boring, and boring is exactly what you want on a production floor at six in the morning, with the first truck already at the gate.
References
- NIST SP 800-82 Rev. 3, Guide to Operational Technology Security: https://csrc.nist.gov/pubs/sp/800/82/r3/final
- CISA, Industrial Control Systems topic page: https://www.cisa.gov/topics/industrial-control-systems
- MITRE ATT&CK T1200, Hardware Additions: https://attack.mitre.org/techniques/T1200/
- ENISA, Critical Information Infrastructures and Services: https://www.enisa.europa.eu/topics/critical-information-infrastructures-and-services
Originally published by Dev.to Security. Aggregated on AIWithGhost for educational purposes β full credit and traffic to the original publisher.

