[
https://issues.apache.org/jira/browse/MAPREDUCE-922?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12846157#action_12846157
]
Steve Loughran commented on MAPREDUCE-922:
------------------------------------------
MAPREDUCE-1603 could provide a way to do this customisation without having to
have logic built into hadoop itself, logic that may date fast and depend upon
the specific workload.
> Automatic configuration of number of map and reduce slots based on available
> resources for dealing with heterogenous clusters
> -----------------------------------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-922
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-922
> Project: Hadoop Map/Reduce
> Issue Type: Sub-task
> Components: tasktracker
> Affects Versions: 0.22.0
> Reporter: Arun C Murthy
>
> Currently the number of map and reduce slots have to be configured manually,
> which is reasonable for homogeneous clusters. However, as the clusters start
> to change over time it becomes rather painful to administer and configure. We
> should start thinking along the direction of auto-magically configuring the
> slots based on available resources on the node such as RAM, CPU, disk etc.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
