You are viewing a plain text version of this content. The canonical link for it is here.
Posted to notifications@iotdb.apache.org by "liaolanyu (Jira)" <ji...@apache.org> on 2023/03/08 14:42:00 UTC
[jira] [Commented] (IOTDB-5522) Last Query performance improvement
[ https://issues.apache.org/jira/browse/IOTDB-5522?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17697935#comment-17697935 ]
liaolanyu commented on IOTDB-5522:
----------------------------------
This doc describes what we think may slow down the last query: [https://apache-iotdb.feishu.cn/docx/GVe2d35aAoVYVhxnueac0hK9nUf]
However, we found current last query might behave wrongly if multiple DataRegions of one timeseries exists on one DataNode.
> Last Query performance improvement
> ----------------------------------
>
> Key: IOTDB-5522
> URL: https://issues.apache.org/jira/browse/IOTDB-5522
> Project: Apache IoTDB
> Issue Type: Improvement
> Components: Core/Query
> Reporter: Yuan Tian
> Assignee: liaolanyu
> Priority: Major
> Attachments: image-2023-02-10-21-23-28-478.png, image-2023-02-16-20-16-11-215.png
>
>
> If we have a lot of data regions and then we execute `select last s1 from root.sg.d1`, this sql may not execute that fast as we expected, because it may need to wait for all data regions' result and then merge it.
> In some user's case, we build a 1C9D cluster, we set dataregion per database to 27. We need 300ms to get the last query result which even slower than MatrixDB.
> !image-2023-02-16-20-16-11-215.png!
--
This message was sent by Atlassian Jira
(v8.20.10#820010)