[ https://issues.apache.org/jira/browse/MAPREDUCE-279?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12996189#comment-12996189 ]
Matei Zaharia commented on MAPREDUCE-279: ----------------------------------------- +1 on decoupling Hadoop MapReduce from the resource management system in a way that allows Hadoop to run on top of other cluster scheduling systems as well. Apart from simplifying experimentation with these types of two-level schedulers, I think this is would be a good thing for the MapReduce project in general as a way to make the project runnable in the maximum variety of environments. For example, there have already been efforts to get Hadoop running on HPC schedulers (e.g. Grid Engine) or Condor, and that would be quite a bit easier with the refactoring that Arun is doing. I imagine that there will be a lot of other work in cluster scheduling in future years, especially as people start running more non-MapReduce applications, so it would be nice to be able to run the Hadoop software stack in these environments. > Map-Reduce 2.0 > -------------- > > Key: MAPREDUCE-279 > URL: https://issues.apache.org/jira/browse/MAPREDUCE-279 > Project: Hadoop Map/Reduce > Issue Type: Improvement > Components: jobtracker, tasktracker > Reporter: Arun C Murthy > Assignee: Arun C Murthy > Fix For: 0.23.0 > > > Re-factor MapReduce into a generic resource scheduler and a per-job, > user-defined component that manages the application execution. -- This message is automatically generated by JIRA. - For more information on JIRA, see: http://www.atlassian.com/software/jira