jackrabbit-oak-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Thomas Mueller (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (OAK-3343) LIRS Cache: investigate concurrency behavior
Date Thu, 10 Sep 2015 14:19:46 GMT

    [ https://issues.apache.org/jira/browse/OAK-3343?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14738804#comment-14738804
] 

Thomas Mueller commented on OAK-3343:
-------------------------------------

http://svn.apache.org/r1702259 (trunk; adding a disabled performance test)
I noticed the Guava cache we use also uses a segment mechanism, same as the LIRS cache. My
tests show that the LIRS cache is almost always faster than the Guava cache (average 30%).
The default settings for the LIRS cache is 16 segments, and for the Guava cache also 16. So
I consider this closed.

> LIRS Cache: investigate concurrency behavior
> --------------------------------------------
>
>                 Key: OAK-3343
>                 URL: https://issues.apache.org/jira/browse/OAK-3343
>             Project: Jackrabbit Oak
>          Issue Type: Bug
>          Components: cache, core
>            Reporter: Thomas Mueller
>            Assignee: Thomas Mueller
>             Fix For: 1.3.7
>
>
> I have a report that the LIRS cache (in combination with the persistent cache) does not
perform well when using 50 threads that concurrently access it. The Guava cache might not
be affected (as much).
> I would like to investigate, with a simple test case, what is the behavior of the LIRS
and Guava cache with highly concurrent use cases, in combination with a cache loader.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message