I am trying to get a feel for a good average shard size.
I have a 6 node cluster
3 nodes are master, data, ingest
3 nodes are data
Each node has 64GB memory and 30GB java heap size
There are about 3100 indices
Each index has 5 shards and 1 replica
Total shard size is 25TB
Average shard size is 685 MB
There are about 36000 shards
We mostly, ingest then do kibana searches, dashboards
Does that sound about right?
Does anyone thing I have too many shards?
Yes, number of shards are on the higher end. In fact average data per shard is quite low (685MB). You can easily increase data per shard to somewhere around 5G, and reduce the total number of shards.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.