mayya-sharipova commented on PR #11743: URL: https://github.com/apache/lucene/pull/11743#issuecomment-1238777470
@msokolov Thanks for your feedback, and good ideas, we can experiment with them. We've discussed within our team (including @jpountz and @jtibshirani) and decided that we still would like to proceed with this change (building graph on indexing) for 9.4 release, as we see benefits outweigh extra memory used. For example, we see significant improvement in refresh time in our Elasticsearch benchmarks, which makes searches on updated data much faster and predictable: <img width="822" alt="image" src="https://user-images.githubusercontent.com/5738841/188762407-0ecaaa9b-71df-4290-83db-630bfd01e5bd.png";> ---- For follow-ups (beyond 9.4 release), we can experiment with : - pre-flush flush (flushing a portion of vectors) - using data with less precision than integers and floats for storing neighbours' info in graph - some other ideas? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
![https://user-images.githubusercontent.com/5738841/188762407-0ecaaa9b-71df-4290-83db-630bfd01e5bd.png"](https://user-images.githubusercontent.com/5738841/188762407-0ecaaa9b-71df-4290-83db-630bfd01e5bd.png"){kind=link}