Comment by vasco
1 day ago
But they'll say their favorite line which is that it can be very disruptive to customers for them to stop the spend, how could they possibly know what to stop in your account?? Always easy to find reasons to explain something when the result is more money for you.
Once they need to stop your spending, the following has to happen
- VMs deleted, along with all the files on them
- S3 buckets emptied
- DNS queries no longer getting answered
- Databases dropped, backups deleted
This is not just something disruptive that can be fixed by spending money again. Even if you stop customer traffic, all these resources are still used and cost money.
For a professional account this is insane. And even for a student account this would be a very bad day. I'd hate for this to happen as a student, because I had a bunch of important stuff there when I was a student.
Why?
>- VMs deleted, along with all the files on them
VMs shut down, deleted after a specified time period (7 days, whatever, the exact time period doesn't matter)
>- S3 buckets emptied
S3 buckets no longer accessible, emptied after a specified time period
>- DNS queries no longer getting answered
Yep, fine
>- Databases dropped, backups deleted
Database access revoked, backup access revoked, deleted after a specified time period.
There is no reason this has to be an immediate deletion. Heck, even if you request they delete something they probably don't do it immediately and just mark it "deleted" until it's cleaned up after time.
If the only choices are "my data gets deleted", or "$1,000 dollar bill", I think most students would choose data deleted.
And the whole "we are nice, we will waive bills for beginners who make mistakes" is in my experience, as a computing professor who has had multiple students hit this, not true. They seem to do waive fairly randomly to my eye, about a third of the time.
> For a professional account this is insane.
At $work there are plenty accounts where this would be acceptable. Annoying perhaps, but nothing that couldn't be recovered in a day would be lost. Reliable cost caps would be especially useful when running stuff with admin permissions and the dependencies don't have airtight supply chain security.
Yes, and I want all of this to happen automatically.
Last time, I had to spent a few hours deleting all the resources one by one, and then double check to be sure nothing was missing.
I did call it didn't I?
None of those things need to happen. (Unless you get billed by the individual DNS query?) The cost of storing things that already exist is known perfectly in advance. It is trivial to include the cost of "maintain everything until the end of the month" in your estimate of how much the customer has already spent. Then, when the spending halt occurs, nothing needs to be deleted because that was already accounted for.