[
https://issues.apache.org/jira/browse/HADOOP-2560?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12644050#action_12644050
]
dhruba borthakur commented on HADOOP-2560:
------------------------------------------
The problem that I am trying to solve is a job that processes a large number of
input files. Each of these files are small. In the existing implementation,
each mapper gets to process one split and each split is very small in size
(limited by the size of the input file). I will investigate the implementation
of MultipleInput.java to see if I can achieve mappers to processes larger
splits.
> Processing multiple input splits per mapper task
> ------------------------------------------------
>
> Key: HADOOP-2560
> URL: https://issues.apache.org/jira/browse/HADOOP-2560
> Project: Hadoop Core
> Issue Type: Bug
> Reporter: Runping Qi
> Assignee: dhruba borthakur
> Attachments: multipleSplitsPerMapper.patch
>
>
> 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.
