Separating index and query hosts


I'm building an ELK2 stack, with two elasticsearch nodes, and was wondering if there is any benefit in configuring our load balancer to direct queries (using kibana) to one node, and indexing (using logstash) to the other? At the moment the load balancer just picks one of the hosts at random for each connection.


Not really, if you query one node then chances are you will be getting data from the other one anyway.

thanks for the reply, would this apply even with the default setup, where each shard has 1 replica, so each node has a full copy of all the data?

Yes, because ES will round robin between a primary and replica shard.