You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by GitBox <gi...@apache.org> on 2022/01/20 11:14:09 UTC
[GitHub] [flink] Aitozi commented on pull request #14893: [FLINK-21321][Runtime/StateBackends] improve RocksDB incremental rescale performance by using deleteRange operator
Aitozi commented on pull request #14893:
URL: https://github.com/apache/flink/pull/14893#issuecomment-1017391939
@lgo Thanks for your contribution, I have applied this patch to our internal version and get a better rescale performance . In my test, the cost of `deleteRange` only at `ms` level .
After we introduce the `deleteRange` api, I think we should tune the `OVERLAP_FRACTION_THRESHOLD` for the initial handle. The current default value is `0.75` is somehow conservative . With the fast `deleteRange` we can choose at least an initial handle . By this we can reduce the cost to `iterator` for one fragmented DB.
--
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: issues-unsubscribe@flink.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org