[
https://issues.apache.org/jira/browse/HDDS-7625?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17651010#comment-17651010
]
Wei-Chiu Chuang commented on HDDS-7625:
---------------------------------------
[~wfps1210] there are other places where we do similar stupid things. For
example, SCMDBCheckpointProvider uses HddsServerUtil to compress SCM RocksDB
dump with gzip.
There's also a public API ReconUtils.createTarFile() which uses tar.gz.
However, it looks like this method is not being used in any production code.
Only in theses. I wonder why they were added in the first place and whether or
not we should remove it.
> Do not compress OM checkpoint
> -----------------------------
>
> Key: HDDS-7625
> URL: https://issues.apache.org/jira/browse/HDDS-7625
> Project: Apache Ozone
> Issue Type: Sub-task
> Reporter: Wei-Chiu Chuang
> Assignee: Chung En Lee
> Priority: Major
> Attachments: Screen Shot 2022-09-27 at 2.53.18 PM.png,
> ozone_om-rhelnn03.cdip.cisco.local.html
>
>
> DBCheckpointServlet
> About 1/5 of CPU time on OM was spent entirely on compressing the rocksdb
> checkpoints with gzip.
> RocksDB data is compressed with Snappy, and so there's no point compressing
> it again with gzip (not to mention gzip throughput is bad)
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
