Im trying to use a custom documentId insertion for every new record.
Will there be any performance impact or any other issues for not using the the automatic id generated by elastic db
Internally generated IDs are trusted to be unique. Externally provided IDs are untrusted and need to be looked up to check that they do not already exist so every write incurs a read.
I've seen someone trying to index all of Twitter and they saw performance degrade as the index grew using tweet IDs as the elasticsearch doc ID. When they switched to elasticsearch-generated IDs the indexing performance remained flat as volumes grew. That was a few years ago and there have been several performance improvements since then but "no read" will always be faster than "a read". Most users however are not trying to index something as big as all Twitter content so it is not a complaint I hear often.
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.