We are working in an Index system with Elasticsearch.¡, and we would like to have our own _id fields, using a unique identifier obtained from the document to be inserted, so Elasticsearch drops a new insert in case this _id is already present in the system.
Roughly a single request unicity check, where the insert just fails to avoid duplication.
My question is, are we loosing performance at Elasticsearch level using our own crafted _id field?
Is the Index keeping up its performance in resharding operations?
Thanks for any advice.