[
https://issues.apache.org/jira/browse/MAPREDUCE-3289?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13139552#comment-13139552
]
Arun C Murthy commented on MAPREDUCE-3289:
------------------------------------------
Todd, have you done any benchmarks with this?
I'm a bit skeptical... one idea might be to keep the whole map-output in the
buffer cache (particularly if it's freed up since the Datanode isn't using as
much of it) to speed up all of the shuffle for the whole job. Also, you might
want to start pre-fetching outputs of other maps of the same job (application)
as an optimization.
Thoughts?
> Make use of fadvise in the NM's shuffle handler
> -----------------------------------------------
>
> Key: MAPREDUCE-3289
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3289
> Project: Hadoop Map/Reduce
> Issue Type: Improvement
> Components: mrv2, nodemanager, performance
> Affects Versions: 0.23.0
> Reporter: Todd Lipcon
> Assignee: Todd Lipcon
> Attachments: mr-3289.txt
>
>
> Using the new NativeIO fadvise functions, we can make the NodeManager
> prefetch map output before it's send over the socket, and drop it out of the
> fs cache once it's been sent (since it's very rare for an output to have to
> be re-sent). This improves IO efficiency and reduces cache pollution.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
