You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@kylin.apache.org by "qiang.wang01" <qi...@hand-china.com> on 2021/09/06 07:52:44 UTC

Kylin3.1.1 Step Build Dimension Dictionary is too slow

Deal all:

   I have met some problem about building cube recently, it will take 60mins+
to Build Dimension Dictionary.

When I cat kylins log ,it show like that:

Loading snapshotTable

![](cid:E5C1441D-AB2B-4A94-A8CD-6F498247FF61@mailmaster)

This Dimension table(actually it is a hive view) has 50w row,

I have merged HDFS's small files,changed the dict to fix_length(8),changed the
rowkey sequence, even make it be the shard by

keyword.but there is no differences.

In my opinion, this step should only be related to the number of
Dimensions,but we just have 50w row.it should not be a high cardinality
problem.

  

So,I need all your helps, Do you have a better solution to make it faster,It
makes me feel despair。orz

![](cid:36D7623A-5AB3-429C-B34C-A281E08A8FF5@mailmaster)