Is there limitation how many indices could I create in ES cluster? and Does
the number of indices affect performance?
I have used DATE as indice for logs from MMO game servers. That give me
chance to remove old data.
But right now, I'm considering use userid as indice, that means there might
be milliion users, as well as million indices created.
Each index comes with a cost and probably having million of indices will require a lot of machines.
Also the cluster state will be a way too big so it could affect cluster stability.
You will probably have at the end of the day a lot of small indices.
I mean: don't do this!
Share indices between users but use routing to make sure all docs for a given user go to the same shard.
Having rolling indices is a good idea though.
Here are my first thoughts about this.
Hope this helps.
--
David
Twitter : @dadoonet / @elasticsearchfr / @scrutmydocs
Is there limitation how many indices could I create in ES cluster? and Does the number of indices affect performance?
I have used DATE as indice for logs from MMO game servers. That give me chance to remove old data.
But right now, I'm considering use userid as indice, that means there might be milliion users, as well as million indices created.
You may use a single index with enough shards for users and use routing for
accessing the shard where a user ID has the docs indexed. See also shard
overallocation
Is there limitation how many indices could I create in ES cluster? and
Does the number of indices affect performance?
I have used DATE as indice for logs from MMO game servers. That give me
chance to remove old data.
But right now, I'm considering use userid as indice, that means there
might be milliion users, as well as million indices created.
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.