[jklowden]

> some systems can never be shut down

Tomshardware could do better reporting. There is no such thing as a computer that can’t fail, or a component that can’t be replaced. Does our reporter think the entire system was installed 25 years ago, and not one component has been replaced since? More likely it’s the ship of Theseus, and not one component is original.

I’ve replaced whole systems without interruption. You build in compatibility, then replace every computer one by one, and phase out use of the compatibility. It’s not rocket surgery.

Technical sites could be superior to the reporting in the general media on technical issues. It doesn’t have to be be stenography.

[bArray]

I think they are just trying to get across how critical this infrastructure is, it can't be powered down for several days and unscheduled maintenance is risky.

The other problem they are up against is that there are not many people around that still understand how it works or what the edge cases are.

Upgrading these large distributed systems can be painful. The NHS tried to upgrade their software, over £10 bn later and they abandoned it [1].

[1] https://www.theguardian.com/society/2013/sep/18/nhs-records-...

[metadat]

Minor nit: There are classes of computers which contain redundant CPUs, PSUs, memory, etc where the components are hot swappable. Very specialized and expensive hardware which is statistically unlikely to ever need a reboot!

Source: My cousin used to sell these systems back in the 90's and 00's.

[stewarts]

IBM Mainframes have these capabilities. But yes, they're quite expensive.