Hi All,
Is there any criteria that should be followed to decide on separating Master nodes and Data nodes?
We usually say when you start heading into double digit node counts, but that's not a hard or fast rule.
It will depend on your use case and requirements.
use case involve ML jobs (approx 100) to run with several rules, alerting, anomaly detections, around 200 GB data ingestion per day with 1 replication.