crunch-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Josh Wills (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CRUNCH-165) Pipelines should automatically use CombineFileInputFormat where input consists of many small files
Date Fri, 22 Mar 2013 16:57:17 GMT

    [ https://issues.apache.org/jira/browse/CRUNCH-165?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13610930#comment-13610930
] 

Josh Wills commented on CRUNCH-165:
-----------------------------------

Yeah, I read through Pig's implementation of this last night. It's basically a bin packing
problem: we get all of the original input splits from a particular input format, and we have
a target size for each of the splits we want to run. If one of the original splits is greater
than the target size, then we run it as is. For the splits that are smaller than the target
size, we try to package them together into a single combined split. Pig's impl is sort of
complicated; I think that doing something simple based on the first-fit decreasing heuristic
[1] will do the trick here. I'm going to make this my project for today.

[1] http://en.wikipedia.org/wiki/Bin_packing_problem
                
> Pipelines should automatically use CombineFileInputFormat where input consists of many
small files
> --------------------------------------------------------------------------------------------------
>
>                 Key: CRUNCH-165
>                 URL: https://issues.apache.org/jira/browse/CRUNCH-165
>             Project: Crunch
>          Issue Type: Improvement
>          Components: Core
>    Affects Versions: 0.4.0
>            Reporter: Dave Beech
>            Assignee: Josh Wills
>         Attachments: CRUNCH-165.patch
>
>
> Hive had a feature introduced in HIVE-74 whereby CombineFileInputFormat would be used
if the input data consisted of many small files, making the resulting mapreduce jobs more
efficient by giving individual mappers more data to process. This would be a nice feature
for Crunch to have, too.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message