Historical Node Replication and load balancing


Curruntly I am running two historical nodes and my data is being ingested using tranquility-kafka . In the task configuration I have set replication factore as 2 so tranqility runs two paraler task in both nodes. Segments are deep stored locally on both machines. Below is my questions.

  1. As coordinator knows about both nodes will it load balance segment loading on both server or it requires S3 for the same.
  2. If S3 required the currently second machine will help only in fault tollrance.
  3. If need concurrency to speed up write process do I need to setup multiple trnquality machine with multi partition segments or anything else I need to do.

Please help. I have already read several discussion but none of them made my points cleared.

any suggestions ?