I'm developing an elasticsearch cluster for a "search" use-case with small real-time ingestion of time series.
Here is my scheduled cluster:
- 3 master-eligible nodes
- 3 data nodes
- 1 ingest node
- 1 coordinator only node
What I would like to know is, how to connect CRUD request to my ES Cluster ? Does the coordinator only will be the only link to user ? For "create", the coordinator will drive my data directly to the ingest node before storage ?
Or should I make difference between writing operations and searching operations (respectively ingest and coordinator) ?