We should aim at better utilization of CPU resources, currently we target 70% commitments on all clusters.
However, the real CPU utilization seems much less than that, like 10%.
The proposition is to overcommit dev clusters and target 100% CPU commitment on prod.
Perhaps dev could be 10 times overcommitted and stg like double