What are the best practices for Cluster Deployment

[ossDataEngineer]

Hello,

I am a Apache Dolphin Scheduler newbie. I would like to know the best practices for setting up ADS in cluster mode. I would like to setup a Zookeeper cluster of 3 VMs with 1 VM designated as ADS master-server and api-server (Data pipeline scheduling/ centralised ADS WebUI rendering) and other 2 VMs as ADS worker-server (which run my data pipelines).

Observation:

1. When I setup ZK cluster, one node is selected as leader and other 2 as followers. Should I run ADS master/api-server on leader only? Or it doesnot matter?
2. Can I create a separate ZK cluster for master/api-server and separate cluster of worker nodes and connect these distinct ZK clusters as one unified ADS cluster?
3. I would like to setup VMs with master/api-server with lower specs to save cost and configure worker-server VMs with more compute/memory/storage

Thanks,
ossDataEngineer