[
https://issues.apache.org/jira/browse/KYLIN-4779?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Zhong Yanghong updated KYLIN-4779:
----------------------------------
Description:
Use InitialCuboidScheduler instead of stats-based TreeCuboidScheduler is not
good for finding best parent cuboid for query performance. Here's the example
for query with source cuboid 0X0200, it has two parent cuboids
* 0X0302 with row count 560K
* 0X0304 with row count 40
If we use InitialCuboidScheduler, it will choose 0X0302 as the target cuboid
for this query. It's obviously better to choose 0X0304.
||Heading 1||Heading 2||
|!0X0302.png|width=400,height=400!|!0X0304.png|width=400,height=400!|
was:Use InitialCuboidScheduler instead of stats-based
> Use TreeCuboidScheduler even when cube planner is not enabled for query
> -----------------------------------------------------------------------
>
> Key: KYLIN-4779
> URL: https://issues.apache.org/jira/browse/KYLIN-4779
> Project: Kylin
> Issue Type: Improvement
> Reporter: Zhong Yanghong
> Assignee: Zhong Yanghong
> Priority: Major
> Attachments: 0X0302.png, 0X0304.png
>
>
> Use InitialCuboidScheduler instead of stats-based TreeCuboidScheduler is not
> good for finding best parent cuboid for query performance. Here's the example
> for query with source cuboid 0X0200, it has two parent cuboids
> * 0X0302 with row count 560K
> * 0X0304 with row count 40
> If we use InitialCuboidScheduler, it will choose 0X0302 as the target cuboid
> for this query. It's obviously better to choose 0X0304.
> ||Heading 1||Heading 2||
> |!0X0302.png|width=400,height=400!|!0X0304.png|width=400,height=400!|
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
