I’ve seen an entire factory shut down for hours because two critical Win10 computers tried and failed to update. It’s never an issue until it becomes one.
Plus a failed update is the whole reason I nuked my C: drive and switched to Manjaro (now running Arch, put down the pitchforks).
Neither of those options were available. It was written by a third-party for some old .NET Framework version, and the server and GUI components were written as a single application. Putting it on a server wasn’t an option either because the application’s GUI was constantly used for the management of assembly machines, and other applications were used for monitoring and administrative stuff.
If you had been there, you’d know why this was a low-priority risk. That place was bleeding from a thousand wounds. At least this had some redundancy, for all it was worth in the end…
(edit) I actually contributed to that software, even though it’s not open-source! I managed to nail down an issue where loading a project file using one locale would result in a crash, but not in others. The .NET stack trace was printed to an XHR response’s payload and I used that to locate a float.ToString() call where CurrentCulture was passed as the cultureInfo instead of InvariantCulture, so depending on the computer’s locale, it would try to parse CSV data either using a decimal dot or a decimal comma. I mailed this to the maintainer and the fix was released within the month.
The operating system is called “Windows Server”. It doesn’t necessarily have to run on a mainframe. It has the regular Windows GUI (with a few differences, the first you’ll notice is “Cntr+Alt+Del to log in”) and can run regular Windows programs.
I work in IT. Windows 10/11 Enterprise is still a bad choice. If it’s a mission-critical system and you must choose Windows, pony up the cash for Windows Server.
The difference between Windows Enterprise and Pro/Home editions is that there are features on it that make my job easier, but it’s still the same shitty operating system under the hood. Windows Server is much more robust and reliable in my experience. Still shit, but slightly less so. It’s designed to run on machines with 24/7 uptime. Windows Enterprise still expects you to regularly restart it for updates and upgrades. That’s alright since we can just set Susan from Finance’s computer to update at 03:00. It’s not okay if that computer controls the entire factory.
You’re not going to have a multiple-day outage. At most, it’ll be a few hours.
If your company isn’t printing out paycheques and payroll advices at least a day before payday, and is instead waiting until the last minute, that represents an organisational failure, not a technical one.
Edit: And don’t forget… the manager can always write the cheques by hand on a paper chequebook.
I’ve seen an entire factory shut down for hours because two critical Win10 computers tried and failed to update. It’s never an issue until it becomes one.
Plus a failed update is the whole reason I nuked my C: drive and switched to Manjaro (now running Arch, put down the pitchforks).
Well, running Windows 10, a consumer user-oriented operating system, to control mission-critical machines is mistake number 1.
This wouldn’t have happened if they had used Windows Server or something actually designed for that task (like Linux!).
Neither of those options were available. It was written by a third-party for some old .NET Framework version, and the server and GUI components were written as a single application. Putting it on a server wasn’t an option either because the application’s GUI was constantly used for the management of assembly machines, and other applications were used for monitoring and administrative stuff.
If you had been there, you’d know why this was a low-priority risk. That place was bleeding from a thousand wounds. At least this had some redundancy, for all it was worth in the end…
(edit) I actually contributed to that software, even though it’s not open-source! I managed to nail down an issue where loading a project file using one locale would result in a crash, but not in others. The .NET stack trace was printed to an XHR response’s payload and I used that to locate a
float.ToString()call whereCurrentCulturewas passed as the cultureInfo instead ofInvariantCulture, so depending on the computer’s locale, it would try to parse CSV data either using a decimal dot or a decimal comma. I mailed this to the maintainer and the fix was released within the month.Windows Server is an option.
The operating system is called “Windows Server”. It doesn’t necessarily have to run on a mainframe. It has the regular Windows GUI (with a few differences, the first you’ll notice is “Cntr+Alt+Del to log in”) and can run regular Windows programs.
Huh… I wonder why there’s versions of it called “Enterprise” then. You might want to talk with Microsoft about their clear mistake. I mean clearly https://www.microsoft.com/en-us/microsoft-365/windows/windows-11-enterprise is in error since you’re correct right?
I work in IT. Windows 10/11 Enterprise is still a bad choice. If it’s a mission-critical system and you must choose Windows, pony up the cash for Windows Server.
The difference between Windows Enterprise and Pro/Home editions is that there are features on it that make my job easier, but it’s still the same shitty operating system under the hood. Windows Server is much more robust and reliable in my experience. Still shit, but slightly less so. It’s designed to run on machines with 24/7 uptime. Windows Enterprise still expects you to regularly restart it for updates and upgrades. That’s alright since we can just set Susan from Finance’s computer to update at 03:00. It’s not okay if that computer controls the entire factory.
I too work in IT… Just because I have some HR users that need to run Quickbooks, I don’t buy them Windows server 2022.
Well, QuickBooks is hardly “mission critical” (usually) although its “server” functionality is so jank that it’s an entirely different discussion.
If you have sufficient outage that paychecks don’t get cut… it will turn mission critical pretty fast.
You’re not going to have a multiple-day outage. At most, it’ll be a few hours.
If your company isn’t printing out paycheques and payroll advices at least a day before payday, and is instead waiting until the last minute, that represents an organisational failure, not a technical one.
Edit: And don’t forget… the manager can always write the cheques by hand on a paper chequebook.
Or even just running the critical machines offline