Thanks in advanced 
I have a production cluster and two non-production clusters (dev/staging) - they are mostly configured the same, but the resources for non production are smaller (pod count, pod cpu request size). We do not set CPU limits. We run in GKE.
Overall nonprod CPU is mostly wasted. I conceptually would like to set CPU request to 0 (or close to it) in non-prod. I am going to start experimenting with that now, but if you have any advice, especially as it relates to HPA configurations when having requests close to zero, that would be appreciated.
Ideally I would like a per env policy like
Development: don’t HPA, just CPU burst
Staging: HPA allowed, but maybe at >100% CPU utilization
Production: HPA, oversized for stability, etc
Ideas welcome. Thanks!