[
https://issues.apache.org/jira/browse/HADOOP-2560?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12557481#action_12557481
]
Doug Cutting commented on HADOOP-2560:
--------------------------------------
> The current system takes full advantage of mapping jobs to nodes dynamically.
Currently we compute and cache the mapping once per job, and then base all
subsequent decisions on that cache. We get ~99% job locality with that
'static' information. Things should be about about the same if we group
things, unless I'm missing something.
> One could perhaps do something like what you suggest dynamically in the JT
> when a TT requests a new job.
That's a possible enhancement. I'm not sure it's required for good
localization, and it would add significant load to the namenode.
> Combining multiple input blocks into one mapper
> -----------------------------------------------
>
> Key: HADOOP-2560
> URL: https://issues.apache.org/jira/browse/HADOOP-2560
> Project: Hadoop
> Issue Type: Bug
> Reporter: Runping Qi
>
> Currently, an input split contains a consecutive chunk of input file, which
> by default, corresponding to a DFS block.
> This may lead to a large number of mapper tasks if the input data is large.
> This leads to the following problems:
> 1. Shuffling cost: since the framework has to move M * R map output segments
> to the nodes running reducers,
> larger M means larger shuffling cost.
> 2. High JVM initialization overhead
> 3. Disk fragmentation: larger number of map output files means lower read
> throughput for accessing them.
> Ideally, you want to keep the number of mappers to no more than 16 times the
> number of nodes in the cluster.
> To achive that, we can increase the input split size. However, if a split
> span over more than one dfs block,
> you lose the data locality scheduling benefits.
> One way to address this problem is to combine multiple input blocks with the
> same rack into one split.
> If in average we combine B blocks into one split, then we will reduce the
> number of mappers by a factor of B.
> Since all the blocks for one mapper share a rack, thus we can benefit from
> rack-aware scheduling.
> Thoughts?
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
