You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@kylin.apache.org by "Julian Hyde (JIRA)" <ji...@apache.org> on 2016/03/11 03:02:41 UTC
[jira] [Commented] (KYLIN-1482) [UMBRELLA] optimize the cuboids to
be computed
[ https://issues.apache.org/jira/browse/KYLIN-1482?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15190347#comment-15190347 ]
Julian Hyde commented on KYLIN-1482:
------------------------------------
Consider using the aggregate optimization algorithm used by Calcite's https://calcite.apache.org/apidocs/org/apache/calcite/materialize/TileSuggester.html.
> [UMBRELLA] optimize the cuboids to be computed
> ----------------------------------------------
>
> Key: KYLIN-1482
> URL: https://issues.apache.org/jira/browse/KYLIN-1482
> Project: Kylin
> Issue Type: Task
> Affects Versions: v1.5
> Reporter: hongbin ma
> Assignee: hongbin ma
>
> As known to many the number of cuboids is exponential to the number of dimensions. Currently in the latest dev branch, Kylin adopt a well designed pruning scheme called "New Aggregation Group" (http://kylin.apache.org/blog/2016/02/18/new-aggregation-group/) that allows user to specify relationships between dimensions for cuboid pruning. The rule-based approach is simple and effective in many situations.
> However the rule-based approach also has its limitations: It will fail to capture potential prunings when the cube designer does not fully understand how the cube will be used, or the nature of the relationship between parent/child cuboid. In this case, we need to provide more flexible tools for cube designer to optimize which cuboids to be computed and which not to.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)