hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chris Dyer (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-2429) The lowest level map-reduce APIs should be byte oriented
Date Mon, 03 Nov 2008 20:16:44 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-2429?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12644797#action_12644797

Chris Dyer commented on HADOOP-2429:

I think this seems like a very prudent suggestion, especially with respect to adding support
for non-Java mappers and reducers that aren't forced to do everything with text.  As for my
understanding of what is typical at Google, serialization is handled in the mappers and reducers
(often using protocol buffers, which incorporate a more efficient encoding for integers, etc.)
and there are indeed many useful cases where complete deserialization is unnecessary and even
where simple identity mapping/reducing is useful.

Is there any chance that this issue is going to be acted on?  As someone who needs to do work
in c++ (without linking in a JVM -- we have memory requirements as well), I've been hesitant
to embrace Hadoop too fully since the restrictions that it imposes on non-Java code are substantial...

> The lowest level map-reduce APIs should be byte oriented
> --------------------------------------------------------
>                 Key: HADOOP-2429
>                 URL: https://issues.apache.org/jira/browse/HADOOP-2429
>             Project: Hadoop Core
>          Issue Type: Improvement
>          Components: mapred
>            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