You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues-all@impala.apache.org by "Amogh Margoor (Jira)" <ji...@apache.org> on 2021/09/29 17:23:00 UTC

[jira] [Created] (IMPALA-10938) Adaptive Broadcast Joins in the Impala

Amogh Margoor created IMPALA-10938:
--------------------------------------

             Summary: Adaptive Broadcast Joins in the Impala
                 Key: IMPALA-10938
                 URL: https://issues.apache.org/jira/browse/IMPALA-10938
             Project: IMPALA
          Issue Type: New Feature
          Components: Backend
            Reporter: Amogh Margoor
            Assignee: Amogh Margoor


Broadcast joins when build side is small can be much faster than the partitioned joins. Whenever cost model figures it out, it would try to enable broadcast join. However due to lack of statistics or due to incorrect estimates it is not always possible to detect small build side. Therefore, we need an adaptive way to figure it out during the runtime and enable the broadcast. There are other systems like Apache Spark and modern dataware houses which do rely on this technique, so it would be good addition to Impala too.

As a part of this, we mayfield  end up defining Adaptive framework which can be extended to other problems like Skew joins etc. 



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-all-unsubscribe@impala.apache.org
For additional commands, e-mail: issues-all-help@impala.apache.org