[
https://issues.apache.org/jira/browse/KYLIN-4335?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
nichunen updated KYLIN-4335:
----------------------------
Fix Version/s: v3.1.0
> Reuse global dictionary from other cube,global domain dict
> ----------------------------------------------------------
>
> Key: KYLIN-4335
> URL: https://issues.apache.org/jira/browse/KYLIN-4335
> Project: Kylin
> Issue Type: New Feature
> Components: Job Engine
> Affects Versions: Future
> Reporter: wangxiaojing
> Assignee: wangxiaojing
> Priority: Major
> Fix For: v3.1.0
>
>
> At present, global dictionary can reuse other columns in the same cube, but
> cannot reuse the dictionaries of other cubes in the cluster.
> It is suggested that not only the global dictionaries of the same cube can
> be reused, but also the global dictionaries of other cubes can be reused,
> which can greatly reduce the build time of dictionaries and avoid the
> repeated build of dictionaries.
> *Global domain dictionary general realization idea:*
>
> {panel:title=Definition}
> the global dictionary column of a cube can rely on the global dictionary
> column of any other cube in the cluster, which is used to achieve the purpose
> of dictionary reuse, reduce repeated construction, and implementation of OneID
> . At present, Kylin has realized the reuse of global dictionary columns
> within the same cube, but it has not realized the reuse of dictionary columns
> across the cube or even across projects.
> {panel}
>
> {panel:title=Step implementation:}
> Step 1, When the user needs to use the global domain dictionary, enable the
> global dictionary option in advanced dictionaries and set reuse
> column(model_name.cube_name.table_name.column);
> Step 2, kylin uses the reused dictionary path to replace the original
> column's dictionary's path where getDictionary(CubeSegment cubeSeg, TblColRef
> col) method is called. The implementation method is similar to the previous
> cube's internal reuse;
> Step 3, When build cube, it mainly involves uploading dictionary data to HDFS
> when build baseid. It is necessary to upload the used project/model/cube and
> dictionary information together.
> {panel}
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
