How to load balance an ES-cluster with 70+nodes and 70+ replica shards?

My es is 70+nodes and there is an index with a pri shard and 70+ replica shards, because there is no new data written, but there are a lot of queries.

but now, only some of them are 80+ cpu and 20+ load (my node is 16c cpu and 64g mem), more than 80% of nodes are 20-60 cpu. i know a large number of requests only access a few nodes. But I have load balance before ES nodes. how to setting ES and make the cold nodes be used more.

Which version of Elasticsearch are you using? How large (size of primary shards) are the indices in question? Apoart from these indices, what other data is held in the cluster and what is the load on this data?

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