the big clouds usually have something you can wrap around alerting on their usage tracking APIs, but that's often very "eventually consistent" and might be dela

the big clouds usually have something you can wrap around alerting on their usage tracking APIs, but that's often very "eventually consistent" and might be delayed by days. Which is understandable in some cases, is still useful for some cases, but e.g. doesn't reliably protect you from "oh crap a broken config launched way to many VMs and now I've spent 10k in a day"