Guidance for Autoscaling replica sets for Temporal Services using `HorizontalPodAutoscaler`

are there any recommendations on ideal proportion of replica sets for different Temporal services?
Let’s say we are setting max replicas for frontend, temporalworker and temporalUI as K.
As there should be more replicas required for persistence related services, should max replicas for history/matching be 2K/4K. How should be go about it?

Could someone provide guidance on this?