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

Posted to issues@flink.apache.org by "Shengkai Fang (Jira)" <ji...@apache.org> on 2021/01/03 09:42:00 UTC

[jira] [Updated] (FLINK-20831) Introduce the in memory topN buffer

     [ https://issues.apache.org/jira/browse/FLINK-20831?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Shengkai Fang updated FLINK-20831:
----------------------------------
    Component/s:     (was: Table SQL / API)
                 Table SQL / Runtime

> Introduce the in memory topN buffer
> -----------------------------------
>
>                 Key: FLINK-20831
>                 URL: https://issues.apache.org/jira/browse/FLINK-20831
>             Project: Flink
>          Issue Type: New Feature
>          Components: Table SQL / Runtime
>    Affects Versions: 1.13.0
>            Reporter: Shengkai Fang
>            Priority: Major
>             Fix For: 1.13.0
>
>
> The buffer is used to cache the record and only keep the topN records. When the invoker decide to emit or snapshot, it can get the sorted records from the buffer. The main benfit of the buffer is that it can reduce the per-record state visit.



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