You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@flink.apache.org by "Anton Kalashnikov (Jira)" <ji...@apache.org> on 2021/10/07 09:36:00 UTC

[jira] [Created] (FLINK-24470) EMA for buffer debloat level

Anton Kalashnikov created FLINK-24470:
-----------------------------------------

             Summary: EMA for buffer debloat level
                 Key: FLINK-24470
                 URL: https://issues.apache.org/jira/browse/FLINK-24470
             Project: Flink
          Issue Type: Improvement
          Components: Runtime / Checkpointing
    Affects Versions: 1.14.0
            Reporter: Anton Kalashnikov
            Assignee: Anton Kalashnikov


Right now, we use EMA for the smoothing calculation of the throughput. But there is an idea to use EMA on the buffer debloat(BufferDebloater) level because it contains more information(old buffer size, current throughput, buffer in use count, etc.) and it looks like it should be used more effectively.

!NB. This is just an idea which makes sense to try. but if the new implementation doesn't have real profit it makes sense to keep the old one.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)