[
https://issues.apache.org/jira/browse/HDFS-1435?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12917768#action_12917768
]
Hairong Kuang commented on HDFS-1435:
-------------------------------------
I thought more about Philip's suggestion. So instead of changing the fsimage
format, I have an option simply compress the whole image file and then when
loading the fsimage, it decompress the image if the image file ends with a
"compression" suffix.
This has a couple of advantages over my original idea:
1. No need to change layout versions;
2. Give admin more flexibility to use existing tools to compress fsimage even
if HDFS is not configured to compress fsimage.
I also did a few experiments with different compression algorithms. I tried
both gzip and LZO with a 13G fsimage, both using the default level of
compression.
Gzip used 13 minutes to compress the 13G fsimage to be 2.3G bytes and
decompression used 2 minutes 47 seconds.
LZO used only 3 minutes to compress the 13G fsimage to be 3G bytes and
decompression used 2 minutes 51 seconds.
This is very promising results. I think fsimage has a lot of duplicate bytes so
it could compress really well. And also it is very obvious that LZO provides
good compression speed and good enough compression quality.
> Provide an option to store fsimage compressed
> ---------------------------------------------
>
> Key: HDFS-1435
> URL: https://issues.apache.org/jira/browse/HDFS-1435
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: name-node
> Affects Versions: 0.22.0
> Reporter: Hairong Kuang
> Assignee: Hairong Kuang
> Fix For: 0.22.0
>
>
> Our HDFS has fsimage as big as 20G bytes. It consumes a lot of network
> bandwidth when secondary NN uploads a new fsimage to primary NN.
> If we could store fsimage compressed, the problem could be greatly alleviated.
> I plan to provide a new configuration hdfs.image.compressed with a default
> value of false. If it is set to be true, fsimage is stored as compressed.
> The fsimage will have a new layout with a new field "compressed" in its
> header, indicating if the namespace is stored compressed or not.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
