We have a cluster of three ElasticSearch nodes with poor reading performances. We need to address why and to improve the reading performances.
Currently, we have tens of millions of documents setted up in 1 primary shard and 2 replicas.
Each shard is about 50 GB.
We're using OpenVPN to connect the nodes.
We are searching for an ElasticSearch expert capable of addressing the bottleneck and solve it.