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

Posted to issues@flink.apache.org by "Yanfei Lei (Jira)" <ji...@apache.org> on 2022/11/21 02:17:00 UTC

[jira] [Commented] (FLINK-28010) Use deleteRange to optimize the clear operation of RocksDBMapState.

    [ https://issues.apache.org/jira/browse/FLINK-28010?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17636391#comment-17636391 ] 

Yanfei Lei commented on FLINK-28010:
------------------------------------

Because the experimental results of [iterator|https://github.com/apache/flink/pull/20405#discussion_r951119238] are not as expected, as [~Ming Li] said, "maybe this is not a good optimization, we moved the cost of delete with Iterator to read, and this cost will persist unless a compaction occurs".  I am going to close this ticket, please reopen this if there are some other optimizations.

> Use deleteRange to optimize the clear operation of RocksDBMapState.
> -------------------------------------------------------------------
>
>                 Key: FLINK-28010
>                 URL: https://issues.apache.org/jira/browse/FLINK-28010
>             Project: Flink
>          Issue Type: Improvement
>          Components: Runtime / State Backends
>            Reporter: ming li
>            Assignee: ming li
>            Priority: Major
>              Labels: pull-request-available
>
> [FLINK-21321|https://issues.apache.org/jira/browse/FLINK-21321] has introduced {{deleteRange}} for fast clipping of Incremental checkpoint, so can the {{clear}} method in {{RocksDBMapState}} be replaced with {{{}deleteRange{}}}?



--
This message was sent by Atlassian Jira
(v8.20.10#820010)