I mean it sounds like a dick move by Backblaze, but comparing it to what OP is describing is downplaying how bad this looks for GCP IMO.
Going through with official contact at GCP, being assured he did everything right and that it was taken care of - and still getting shut down. In my book that's just inexcusable - if you can't count on that kind of support to work then how can you rely on that platform for anything ?
They're cheap and reliable, that generally comes at the risk of early termination. I don't think you can have a cheap, reliable, unlimited service; pick two at most.
If everything is metered it seems kind of odd to complain about a lot of activity. Since it's literally the thing you make money with.
Minor nitpick, but this is categorically not true. As a customer, one who proactively worked with AWS AMs, SAs, and S3 engineers directly before making the change, I still took down S3 by just changing how new files were stored in our S3 bucket. Yes, this S3 bucket was at the time the largest bucket (based on number of files), but still not “simply unbreakable”.
Backblaze has previously locked our account on a saturday.
They claim we were putting too much pressure on their cluster.
They could've locked the 1! token that was causing 90% of the load, but they locked the whole account which was used by 10 different apps for a veriety of purposes causing unecessary downtime in some production workloads.
Also the workload has been quite consistent for days/weeks, so they could've detected it earlier and/or address it in a better way.