What influences the number of inverted indexes?
Is it one per node or is it one per index or something else? Is it a good assumption that the less inverted indexes I have the better is?
Couldn't google the answer so please help.
Hi Zoltan,
as you noticed, the meaning of "index" is a bit overloaded in Elasticsearch. An Elasticsearch index is a logical unit that groups together related data. It consists of one or more "shards" which can be distributed across one or more "nodes" (basically running ES processes). Each shard itself is an individual Lucene index which can be conceptually be thought of as an inverted index, but in reality itself consists of many "segments"... So you see it's a bit like a nesting doll really.
Here's for example a blog post that presents this nesting from the bottom up. Having said that, having few inverted indices is generally not that important, under certain circumstances it might help, but typically you shouldn't worry about the number of inverted indices (meaning segments on the lucene level) since that is managed by Lucene internally.
This is
Thank you for the clarification, it was very helpful to me.