[ses1984]

Your own instances can have an unexpected CPU spike.

Even if you're running on bare metal I find it hard to believe you don't have a layer with short timeouts between your front and backend.

[fxtentacle]

Why would I? I have redundant 1GBit LAN cables between front end, back end, and database servers.

[ses1984]

Because it's bad ux for your users to see a spinning loading icon forever.

[fxtentacle]

And a timeout error would be better?

[ses1984]

In my experience, yes, a lot better.