Performance issues with _search query

We are running ES 6.5 cluster with 10 data node and 3 front node on i3.2xlarge.
We have daily indices with records-2019-08-10, records-2019-08-11 etc
Each index has 50 million records with 20GB daily data size.

the index is having 2 shards and 2 replicas

We are running a _search query with indices pattern records-2019-08-1* (means the search is done for 10 days)
The query is as follows {"size":50,"timeout":"60s", "_source":["field1","field2","field3"],"query":{"bool":{"filter":[{"match":{"field1":{"query":"xxx"}}},{"range":{"field2":{"from":234}}}]}},"sort":[{"field2":{"order":"asc"}}]}
While running the load test I am hitting the front nodes and the test results return max of 40 requests/sec with average latency 5 secs .
There is no active write done to the above-mentioned indices.
The results we are getting are way less than we expect our production taking too.
I am looking for ways to optimize the performance of the index.
We want to scale it to 500-600 RPS.
Let me know if there are suggestions

What is the average size of your documents? What does the match query string look like?

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