My understanding of the docs (any a quick convo with gemini) is that in regional clusters, the pods are replicated across the node pools, and then a loadbalancer balances the traffic across the pods. Meaning that, if I deploy a hello-world pod, I will essentially deploy multiple replicas, thus removing the need for me to create the replicas myself.

Is my understanding correct?

If my understanding is correct:

How would HA deployments such as elasticsearch or postgres-ha work? How would they share state if they don’t know about each other? How would they do leader-election or stuff like that?

If my understanding is incorrect:

• How is that different that multi-zonal regions? Is it just the replication of the control-plane?
• Do the control-planes work together?