rmi performance differences

[Kevin Pfaff <ke...@byteconsult.de>]

Hello -
 
I have a big problem with the performance with jcr over rmi on different pcs. We have the same jackrabbit-standalone (version 2.2.7) running on 3 different pcs. Two of them are the same pcs (hardware) and one is more powerful. The access speed on the more powerful one and one of the other two is much slower (1-2 accesses per second) than on the third one (more than twenty). We use the exact same jcr-standalone (version 2.2.7), operating system (windows 7 sp1), Eclipse version (indigo),  jdk (1.6.0_26) and program (the one we are developing). The servers (jackrabbit-standalone) are running on localhost on all three pcs each. The programs are running on tomcat 6 in eclipse and connect to the jackrabbit-standalone on the same pc over rmi.
Anybody some suggestions?
 
Thanks,
Kevin 
 


Behalten Sie die Zukunft von Marketing und IT im Blick. Abonnieren Sie unseren Newsletter unter http://newsletter.byteconsult.de

Re: rmi performance differences

[Jukka Zitting <ju...@gmail.com>]

Hi,

On Thu, Sep 8, 2011 at 1:37 PM, Kevin Pfaff <ke...@byteconsult.de> wrote:
> I tried the same code as an embedded reposatory of jackrabbit and the
> time for the code was 0ms.

OK, it then sounds clear that the issue is with RMI or network
configuration. For example, is it possible that the two slow machines
already have a separate rmiregistry process running? You might also
want to enable extra RMI logging [1] to get a better picture of what
happens during a remote method call.

[1] http://download.oracle.com/javase/1.5.0/docs/guide/rmi/logging.html

BR,

Jukka Zitting

AW: rmi performance differences

[Kevin Pfaff <ke...@byteconsult.de>]

Hi,
On Thu, Sep 8, 2011 at 13:32 PM, Jukka Zitting  <ju...@gmail.com> wrote:
> It would be useful if you could determine whether this is a remote access problem or a repository configuration issue.
> For example, do you still see the problem if you use Jackrabbit as an embedded repository instead of using it remotely over RMI?

I tried the same code as an embedded reposatory of jackrabbit and the time for the code was 0ms.

Thanks,
Kevin



Behalten Sie die Zukunft von Marketing und IT im Blick. Abonnieren Sie unseren Newsletter unter http://newsletter.byteconsult.de

Re: rmi performance differences

[Jukka Zitting <ju...@gmail.com>]

Hi,

On Thu, Sep 8, 2011 at 12:53 PM, Kevin Pfaff <ke...@byteconsult.de> wrote:
> The repository comes from Spring IDE with autowire and the setting for the
> repository are the same for all 3 pcs. The fast pc needs for this little code
> about 5 milliseconds the other two need over 500 ms. We even tried to
> connect to the repository of the "fast access pc" over network, but still
> the same results, it is only fast for itself. Some suggestions?

It would be useful if you could determine whether this is a remote
access problem or a repository configuration issue. For example, do
you still see the problem if you use Jackrabbit as an embedded
repository instead of using it remotely over RMI?

BR,

Jukka Zitting

AW: rmi performance differences

[Kevin Pfaff <ke...@byteconsult.de>]

I only started the jackrabbit-standalone and worked with the generated repisatory.xml.

I now can pinpoint the problem after some testing. The reason for the long access times comes from the login. I stopped time for this code:

		Credentials credentials = new SimpleCredentials("Test", "test".toCharArray());
		this.repository.login(credentials);

The repository comes from Spring IDE with autowire and the setting for the repository are the same for all 3 pcs. The fast pc needs for this little code about 5 milliseconds the other two need over 500 ms. We even tried to connect to the repository of the "fast access pc" over network, but still the same results, it is only fast for itself. Some suggestions?

Regards,
Kevin


Behalten Sie die Zukunft von Marketing und IT im Blick. Abonnieren Sie unseren Newsletter unter http://newsletter.byteconsult.de

Re: rmi performance differences

[maikhorma <MH...@NBME.org>]

What persistence are you using? Are these clustered?  

--
View this message in context: http://jackrabbit.510166.n4.nabble.com/rmi-performance-differences-tp3792785p3797262.html
Sent from the Jackrabbit - Users mailing list archive at Nabble.com.