Elastic search coordinating-node

Question

we are new to elasticsearch and beginning to set-up a coordination node for our UI client to query the index. didn't really understand the difference between master node and coordination node. does coordination has to be scaled up separately based in the site traffic? will other nodes share the load?

Answer 1

The master node is responsible for managing the cluster topology. It neither indexes data nor participates in search tasks.

The data nodes are the real work horses of your ES cluster and are responsible for indexing data and running searches/aggregations.

Coordinating nodes (formerly called "client nodes") are some kind of load balancers within your ES cluster. They are optional and if you don't have any coordinating nodes, your data nodes will be the coordinating nodes. They don't index data but their main job is to distribute search tasks to the relevant data nodes (which they know where to find thanks to the master node) and gather all the results before aggregating them and returning them to the client application.

So depending on your cluster size, amount of data and SLA requirements, you might need to spawn one or more coordinating nodes in order to properly serve your clients. Without any real numbers, it is hard to advise anything at this point, but the above describes how each kind of node works.

If you're just beginning and don't have much data, you don't need any dedicated coordinating node, a simple data node is perfectly fine.