I trying to use Tranquility+Indexing service to consume kafka messages. but in Tranquility doc(https://github.com/druid-io/tranquility/blob/master/docs/overview.md), I found that “You don’t need any realtime nodes, since Tranquility uses the indexing service for all of its ingestion needs.”
Does it means I shouldn’t setup any realtime nodes in cluster? if so, does the realtime data can be query before it handoff to the historical nodes? and where the realtime data be query?
I’m not clear about this matter, please help me to understand the relationship between realtime node and index serivce