[jira] [Commented] (KYLIN-2409) Performance tunning for in-mem cubing

["Shaofeng SHI (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/KYLIN-2409?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15956218#comment-15956218 ] 

Shaofeng SHI commented on KYLIN-2409:
-------------------------------------

By analysis the profiling report from JMC, only two findings:

1) TrieDictionaryForest repeatedly calculate something, this has been fixed by xiefan.
2) The mapper works in a multiple thread way, which over-subscribe CPU and impacts other mappers; by changing to single thread, the overall performance get a small enhance. 

> Performance tunning for in-mem cubing
> -------------------------------------
>
>                 Key: KYLIN-2409
>                 URL: https://issues.apache.org/jira/browse/KYLIN-2409
>             Project: Kylin
>          Issue Type: Improvement
>          Components: Job Engine
>            Reporter: Shaofeng SHI
>            Assignee: Shaofeng SHI
>             Fix For: v2.0.0
>
>
> Some time we observed the in-mem cubing is slow. It deserves a profiling to see whether there are bottlenecks.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)