Search performance issue on Elasticsearch cluster

In an 14 node, RAID 5 based Elasticsearch cluster, the search performance does not improve no matter how simple the full-text query is.

There are two indices: one is 30 shard, 1 replica, 1.1tb and 110M document size. The other is 200 shard, 1 replica, 8.4tb and 663M document size. In both of them, the average query time does not seem to be lower than 700 ms.

The query is basically a multi_match query in few different fields.

What could be the reason?

What does disk I/O and iowait look like when you are running queries? What dors CPU usage look like? Is there anytjong about long or frequent GC in the logs? Which version are you using?

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.