We have a bunch of this shit still around for instruments. Every year we tell them to budget for replacement. Every year someone comes screaming that their 2000 or XP box isn’t booting. Sorry, hardware failure… And we don’t keep anything around that can support it anymore. If it was that important to production you should have replaced it a long time ago.
I used to work in a factory; we had two machines that were designed in the 80’s. Both had the “upgrade” package purchased some time back; they were running the latest WinXP (home) on embedded computers.
The company went out of business sometime in the early 00’s; the replacement from a competitor was $7M USD (but only in the states) ~$15M NZ to ship and install.
So for ~$30M we could replace perfectly working equipment with something that did the same job…because of an old computer. We would have achieved no extra production, or better quality. So the decision was, every year, keep it going as is.
It sometimes is not as simple as “you should have upgraded already”.
The contingency plan I helped implement was to virtualise the application in a VM on the main server; run the comms to the embedded system using a serial to ethernet converter. I tested it a lot and found that it worked very well…never put it into production as it was “backup only” but it was better than hoping it wouldn’t fail. The embedded systems themselves were relatively simple with replacement components that were not too difficult to get hold of.
We have a bunch of this shit still around for instruments. Every year we tell them to budget for replacement. Every year someone comes screaming that their 2000 or XP box isn’t booting. Sorry, hardware failure… And we don’t keep anything around that can support it anymore. If it was that important to production you should have replaced it a long time ago.
I would like to add some context.
I used to work in a factory; we had two machines that were designed in the 80’s. Both had the “upgrade” package purchased some time back; they were running the latest WinXP (home) on embedded computers.
The company went out of business sometime in the early 00’s; the replacement from a competitor was $7M USD (but only in the states) ~$15M NZ to ship and install.
So for ~$30M we could replace perfectly working equipment with something that did the same job…because of an old computer. We would have achieved no extra production, or better quality. So the decision was, every year, keep it going as is.
It sometimes is not as simple as “you should have upgraded already”.
The contingency plan I helped implement was to virtualise the application in a VM on the main server; run the comms to the embedded system using a serial to ethernet converter. I tested it a lot and found that it worked very well…never put it into production as it was “backup only” but it was better than hoping it wouldn’t fail. The embedded systems themselves were relatively simple with replacement components that were not too difficult to get hold of.