benwtrent commented on PR #15224: URL: https://github.com/apache/lucene/pull/15224#issuecomment-4914612480
I am gonna close this. Honestly, all benchmarking found the tradeoffs here to not be worth it. I think for anything direct IO focused that wants to handle throughput, we need to use some native things...or just stick with mmap and be ok with the page faults. FWIW, in benchmarking we found just doing pre-fetch with MMAP and multiple threads reached IO saturation, while this PR never really did, and consequently read throughput was always worse :( -- 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]
