Once we were having trouble with GKE hosted in Asia, it was causing out business a major outage and it wasn't something which I had the power to fix, from memory, half way through a cluster upgrade, Google ran out of compute so the upgrade was stuck half and the control plane ended up in a bad state and some how this impacted the networking (it shouldn't but it happened). I was unable to provision a new cluster due to the lack of capacity so we were stuck.

This wasn't the first problem we'd had either.

There was absolutely no one to call, no one to even alert to warn other customers, the status pages were all green.

Instead of bothering with Google, I just opened an account on AWS and migrated whole stack to AWS in ~ 3 hours, pointed DNS at the new load balancers and we never went back and continued doing business without issue for as long as I can remember.