[
https://issues.apache.org/jira/browse/HDDS-9232?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17765394#comment-17765394
]
Mladjan Gadzic commented on HDDS-9232:
--------------------------------------
[~szetszwo], I run write performance again using 1.3.0 version as baseline and
more recent master branch. When using properties which values are set to 0ms,
it runs faster by the order of magnitute. I am wondering, why these properties
aren't used by default? Is there a drawback to this?
Here are the results:
{code:java}
# (recent master) - fso - no properties
Total Execution time: 00:26:20,628
# (recent master) - fso - properties set to 0ms
Total Execution time: 00:04:25,622
# 1.3.0 - fso - no properties
Total Execution time: 00:03:45,593
# 1.3.0 - fso - properties set to 0ms
Total Execution time: 00:03:50,571
{code}
> Write performance degradation
> -----------------------------
>
> Key: HDDS-9232
> URL: https://issues.apache.org/jira/browse/HDDS-9232
> Project: Apache Ozone
> Issue Type: Bug
> Affects Versions: 1.3.0
> Reporter: Mladjan Gadzic
> Priority: Blocker
> Labels: ozone-performance
>
> It is observed on real cluster using unsecure Ozone HA (3 OMs, 3 SCMs, 3DNs)
> that write performance degraded significantly after Ratis upgrade to 2.5.1
> (https://issues.apache.org/jira/browse/HDDS-8382). At the time of writing
> this ticket, it takes master branch ~25mins for 100k keys to be written,
> while it took 1.3.0 release ~2mins for the same number of keys.
> Test was done using single freon client on OM node. Command
> {code:java}
> ozone freon rk --bufferSize 4096 --factor THREE --keySize 10240
> --numOfVolumes 1 --numOfBuckets 1 --numOfKeys 100000 --numOfThreads 10
> --replication-type RATIS {code}
> was used for that purpose.
> Commit before HDDS-8382, HDDS-8550 (sha
> 290e5f2ba1826a2084c1745e458872dc76c7ae91) did not exhibit degradation in
> write performance.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
