hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Arun C Murthy (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-1194) map output should not do block level compression
Date Tue, 03 Apr 2007 03:17:32 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-1194?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12486213
] 

Arun C Murthy commented on HADOOP-1194:
---------------------------------------

Runping, 

'mapred.compress.map.output' is the boolean which controls whether map outputs are compressed.
If set to true (@see MapTask.java:279) the 'mapred.output.compression.type' can be used to
specify RECORD/BLOCK - this defaults to SequenceFile's compression-type i.e. 'io.seqfile.compression.type'
which defaults to RECORD compression. (@see JobConf.getMapOutputCompressionType).

Does this help? Or do you propose we change JobConf.getMapOutputCompressionType to return
RECORD if 'mapred.output.compression.type' is unset i.e. do not use io.seqfile.compression.type?
Thanks for the clarification...

> map output should not do block level compression
> ------------------------------------------------
>
>                 Key: HADOOP-1194
>                 URL: https://issues.apache.org/jira/browse/HADOOP-1194
>             Project: Hadoop
>          Issue Type: Bug
>          Components: mapred
>    Affects Versions: 0.12.2
>            Reporter: Runping Qi
>
> If the user sets to compress the map output, the compression style should be record level,
not block level, since using block level compression for map outputs causes performance degragation
significantly.

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


Mime
View raw message