You are viewing a plain text version of this content. The canonical link for it is here.
Posted to mapreduce-dev@hadoop.apache.org by "Jothi Padmanabhan (JIRA)" <ji...@apache.org> on 2009/09/11 06:26:57 UTC
[jira] Resolved: (MAPREDUCE-240) Improve the shuffle phase by using
the "connection: keep-alive" and doing batch transfers of files
[ https://issues.apache.org/jira/browse/MAPREDUCE-240?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jothi Padmanabhan resolved MAPREDUCE-240.
-----------------------------------------
Resolution: Duplicate
MAPREDUCE-318 incorporated this
> Improve the shuffle phase by using the "connection: keep-alive" and doing batch transfers of files
> --------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-240
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-240
> Project: Hadoop Map/Reduce
> Issue Type: Improvement
> Reporter: Devaraj Das
> Assignee: Jothi Padmanabhan
> Attachments: hadoop-1338-v1.patch, hadoop-1338-v2.patch
>
>
> We should do transfers of map outputs at the granularity of *total-bytes-transferred* rather than the current way of transferring a single file and then closing the connection to the server. A single TaskTracker might have a couple of map output files for a given reduce, and we should transfer multiple of them (upto a certain total size) in a single connection to the TaskTracker. Using HTTP-1.1's keep-alive connection would help since it would keep the connection open for more than one file transfer. We should limit the transfers to a certain size so that we don't hold up a jetty thread indefinitely (and cause timeouts for other clients).
> Overall, this should give us improved performance.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.