[
https://issues.apache.org/jira/browse/KYLIN-747?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14979709#comment-14979709
]
Vadim Semenov commented on KYLIN-747:
-------------------------------------
Tested changes from this commit:
https://github.com/apache/incubator-kylin/commit/f96600e89a5a2c0e533cea86ad6a73b9451bcddc
Before the changes, queries on partitioned cubes with filters where only some
partitions have relevant data (including partition columns) resulted in
scanning every partition, after the change only relevant partitions are scanned.
Great enhancement, thank you [[email protected]].
> bad query performance when IN clause contains a value doesn't exist in the
> dictionary
> -------------------------------------------------------------------------------------
>
> Key: KYLIN-747
> URL: https://issues.apache.org/jira/browse/KYLIN-747
> Project: Kylin
> Issue Type: Bug
> Reporter: ZhouQianhao
> Assignee: liyang
> Fix For: v2.0, v1.1
>
>
> For example, we had a query of 39 x values in where clause and there is one
> x value not in cube, which yielded the following result:
> Duration: 60.947
> Cube Names: [olap]
> Total scan count: 2524898
> Result row count: 39
> (The log shows "Can't translate value xxx to dictionary ID, roundingFlag 0.
> Using default value \xFF")
> And we excluded the x value that is not in cube and re-run the query and got
> the another result:
> Duration: 2.477
> Cube Names: [olap]
> Total scan count: 96543
> Result row count: 38
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
