camel-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Fabien Chaillou (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CAMEL-6264) Camel 2.10 performance regression with added statistics in BaseTypeConverterRegistry
Date Fri, 19 Apr 2013 17:05:17 GMT

    [ https://issues.apache.org/jira/browse/CAMEL-6264?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13636606#comment-13636606
] 

Fabien Chaillou commented on CAMEL-6264:
----------------------------------------

Hello,

Sorry i did not take the time to answer earlier.
Thank you very much for your fix.
For the record, my test was on a Linux 3.5 and we had around 50 threads
(which is too much anyway and we reduced it to 10 threads in the process)

I will soon take some time to retry our test and i'll let you know the
result after the fix.
Regards
Fabien Chaillou,
Triton Digital


2013/4/14 Claus Ibsen (JIRA) <jira@apache.org>


                
> Camel 2.10 performance regression with added statistics in BaseTypeConverterRegistry
> ------------------------------------------------------------------------------------
>
>                 Key: CAMEL-6264
>                 URL: https://issues.apache.org/jira/browse/CAMEL-6264
>             Project: Camel
>          Issue Type: Improvement
>          Components: camel-core
>    Affects Versions: 2.10.3, 2.11.0
>            Reporter: Fabien Chaillou
>            Assignee: Claus Ibsen
>              Labels: performance
>             Fix For: 2.10.5, 2.11.0
>
>
> We updated one of our applications from camel 2.7.3 to camel 2.10.3.
> We have a camel route that takes its input from a servlet and we saw a 10% performance
drop after the update.
> With more tests using yourkit, it appears that we are spending a lot more time in BaseTypeConverterRegistry#convertTo
method than we used to (from 10ms to 3200ms in a test case).
> The only difference between the two version is the added statistics and the AtomicLong
that must cause contention on our route that uses multiple threads behind a seda.
> To improve the issue on our side, we worked-around to avoid having a type converter in
our hotspots (Which was mainly the DefaultHttpBinding#writeResponse from camel-http), but
i think a solution should be find (maybe make these statistics optionnal).
> Regards,
> Fabien

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message