Performance considerations on uid generation

I am trying to evaluate the best practices for ID generation in terms of index/query time performance and also storage; and stumbled upon this PR

  • It sounds like there will be some benefits if I send base64 generated IDs?
  • And also the length of the ID might also have some performance overhead?
  • Having a common prefix might improve compression? turns out I didn't finish reading https://github.com/elastic/elasticsearch/issues/18209

I would like to know if these are still valid in 7.x and is there any other documentation/blog posts where this is explained conclusively about the benefits?

Have a look at the following resources:

https://www.elastic.co/webinars/optimizing-storage-efficiency-in-elasticsearch

1 Like

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.