You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Apache Spark (JIRA)" <ji...@apache.org> on 2014/09/20 10:25:33 UTC
[jira] [Commented] (SPARK-3613) Don't record the size of each
shuffle block for large jobs
[ https://issues.apache.org/jira/browse/SPARK-3613?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14141870#comment-14141870 ]
Apache Spark commented on SPARK-3613:
-------------------------------------
User 'rxin' has created a pull request for this issue:
https://github.com/apache/spark/pull/2470
> Don't record the size of each shuffle block for large jobs
> ----------------------------------------------------------
>
> Key: SPARK-3613
> URL: https://issues.apache.org/jira/browse/SPARK-3613
> Project: Spark
> Issue Type: Improvement
> Components: Shuffle, Spark Core
> Reporter: Reynold Xin
> Assignee: Reynold Xin
>
> MapStatus saves the size of each block (1 byte per block) for a particular map task. This actually means the shuffle metadata is O(M*R), where M = num maps and R = num reduces.
> If M is greater than a certain size, we should probably just send an average size instead of a whole array.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org