Even after reading the documentation, I'm still not able to understand the utility of using "min-idle-instances" if idle instance is billed the same way as a running instance.
So for example rather than doing "min_instances: 1, min-idle-instances: 1" lets do "min_instances: 2" if it costs the same, and if the purpose is to reduce pending latency, because a "running" instance will be always ready to serve.
Am I missing something ?
Regarding pricing, you're absolutely right.
App Engine charges per instance, whether active or inactive.
min_idle_instance
is not intended for cost optimization, but rather for reducing latency due to cold starts. It's primarily used for fine-tuning your application to handle unexpected high demand efficiently.