Hacker News new | past | comments | ask | show | jobs | submit login

Cloud Run PM here: I'm sorry for the bad experience the customer shared in this article, we could certainly do better with bill management.

We pick 1,000 as a default value for "maxScale", this can be considered high for some users, but low for users who expect infinite scaling from the service and start with a load test to evaluate it.




> We pick 1,000 as a default value for "maxScale", this can be considered high for some users, but low for users who expect infinite scaling from the service and start with a load test to evaluate it.

That seems absurd to me.

I think it makes much more sense to put the onus on the sophisticated customer to increase their maxScale to an unusual value. Users who "expect infinite scaling...and start with a load test" are sophisticated users.

E.g. set maxScale low, like 2 or 4. The sophisticated customer would recognize their oversight quickly. Click-click, fixed, restart test.

Effectively 100% of less-sophisticated customers will not need enormous scale on day 1. Customers with whom you do not have an existing billing relationship in the 10s of thousands of dollars per cycle will almost certainly not want it.

I'd consider that level of overspecification to be a strong anti-pattern.



Given that this is a common problem, and one that can bankrupt individuals or their businesses, when is AWS going to implement spending caps that are easy to set up for new developers or business owners?


Still the really expensive thing here was the datastore reads. Cpu time was only 10% of the bill




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: