[
https://issues.apache.org/jira/browse/HADOOP-12794?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15147940#comment-15147940
]
Ravi Mutyala commented on HADOOP-12794:
---------------------------------------
Thanks Junping. I am working on adding the Tests and checkstyle and will upload
a new patch soon.
Regarding adding BEST_BALANCE for compression level 4 (not 6 which is the
default), that is a good option but would like more inputs to confirm that our
internal tests about level 4 being best balance is true with other datasets. We
could name the levels as
BEST_SPEED(1),TWO(2), THREE(3),BEST_BALANCE(4),FIVE(5),DEFAULT(6),SEVEN(7),
EIGHT(8),BEST_COMPRESSION(9).
The reason for leaving numbers (TWO,THREE.. etc) would be give flexibility of
using all compression levels available in zlib in compressioncodec.
> Support additional compression levels for GzipCodec
> ---------------------------------------------------
>
> Key: HADOOP-12794
> URL: https://issues.apache.org/jira/browse/HADOOP-12794
> Project: Hadoop Common
> Issue Type: Improvement
> Components: io
> Affects Versions: 2.7.2
> Reporter: Ravi Mutyala
> Assignee: Ravi Mutyala
> Fix For: 2.7.3
>
> Attachments: HADOOP-12794.0001.patch
>
>
> gzip supports compression levels 1-9. Compression level 4 seems to give best
> compression per CPU time in some of our tests. Right now ZlibCompressor that
> is used by GzipCodec only supports levels 1,9 and six (default).
> Adding all the compression levels that are supported by native ZlibCompressor
> can provide more options to tweak compression levels.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
