impala-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Tim Armstrong (JIRA)" <j...@apache.org>
Subject [jira] [Created] (IMPALA-5120) Consider defaulting to partitioned join when no stats are available.
Date Mon, 27 Mar 2017 23:29:41 GMT
Tim Armstrong created IMPALA-5120:
-------------------------------------

             Summary: Consider defaulting to partitioned join when no stats are available.
                 Key: IMPALA-5120
                 URL: https://issues.apache.org/jira/browse/IMPALA-5120
             Project: IMPALA
          Issue Type: Improvement
          Components: Frontend
    Affects Versions: Impala 2.9.0
            Reporter: Tim Armstrong


We currently default to broadcast join when no stats are available, since the code estimates
are both MAX_LONG and in the case of equal costs, broadcast wins. We should consider making
partitioned join the default because it will use less memory.

The code is here: https://github.com/apache/incubator-impala/blob/master/fe/src/main/java/org/apache/impala/planner/DistributedPlanner.java#L509
{code}
    && (node.getDistributionModeHint() == DistributionMode.BROADCAST
            || (node.getDistributionModeHint() != DistributionMode.PARTITIONED
&& broadcastCost <= partitionCost)))
{code}



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

Mime
View raw message