hadoop-mapreduce-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Tom White (JIRA)" <j...@apache.org>
Subject [jira] Commented: (MAPREDUCE-326) The lowest level map-reduce APIs should be byte oriented
Date Fri, 05 Feb 2010 22:52:28 GMT

    [ https://issues.apache.org/jira/browse/MAPREDUCE-326?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12830346#action_12830346

Tom White commented on MAPREDUCE-326:

There are some great ideas in this thread that I would love to see implemented, but I'd like
to focus on a few simple steps we can take to building a low-level binary API.

In some sense we almost have a binary API tucked away in Task/MapTask/ReduceTask - it's just
not possible to get at for framework developers. So we refactor the Task, MapTask, and ReduceTask
classes to promote their nested classes to top level-classes. This would be done in conjunction
with moving them to o.a.h.mapreduce.task and subpackages, and marking them using the private
annotation.  Then we can clean up their interfaces to make things more uniform (hiding the
old and new APIs behind a common API). I think we can do this work in stages, and it will
improve the modularity of this area of the codebase without sacrificing efficiency or compatibility.


> The lowest level map-reduce APIs should be byte oriented
> --------------------------------------------------------
>                 Key: MAPREDUCE-326
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-326
>             Project: Hadoop Map/Reduce
>          Issue Type: Improvement
>            Reporter: eric baldeschwieler
> As discussed here:
> https://issues.apache.org/jira/browse/HADOOP-1986#action_12551237
> The templates, serializers and other complexities that allow map-reduce to use arbitrary
types complicate the design and lead to lots of object creates and other overhead that a byte
oriented design would not suffer.  I believe the lowest level implementation of hadoop map-reduce
should have byte string oriented APIs (for keys and values).  This API would be more performant,
simpler and more easily cross language.
> The existing API could be maintained as a thin layer on top of the leaner API.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message