[jira] [Commented] (KYLIN-3116) Speed up load hive metadata

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

    [ https://issues.apache.org/jira/browse/KYLIN-3116?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16294904#comment-16294904 ] 

Shaofeng SHI commented on KYLIN-3116:
-------------------------------------

Currently, the job will collect statistics over the whole table; when the source table is big, the job is expensive. You can uncheck that option when load a hive table, this step is optional.

There is another task I remember is, allowing user to select one or multiple partitions to run the statistics, that will ease the burden.

> Speed up load hive metadata
> ---------------------------
>
>                 Key: KYLIN-3116
>                 URL: https://issues.apache.org/jira/browse/KYLIN-3116
>             Project: Kylin
>          Issue Type: Bug
>          Components: Metadata
>    Affects Versions: v2.2.0
>         Environment: Centos 7, ambari, HDP 2.6
>            Reporter: vu thanh dat
>            Assignee: Shaofeng SHI
>              Labels: build
>             Fix For: 1.x-HBase1.1.3, v2.2.0
>
>         Attachments: mapreduce.bmp
>
>
> Hi all,
> im using Kylin to speed up query HDFS.
> Which 2.5 billions records per day. When i'm using Kylin and load hive table, it make a mapreduce on my server, 2969 map and 1 reduce with only 1 table. My datasource about 10 tables. And it so slow. I attached an image.
> So, can i speed up process "Load hive table" to create model and cubes so quickly?
> Exam use spark (because mapreduce so slow)?
> Best regards, Thanks all



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)