[seam-dev] JNDI Lookups, and UserTransation lookup breakdown

Jay Balunas tech4j at gmail.com
Thu Oct 2 16:19:55 EDT 2008


The seam-gen test went about the same.  I used the same seam-gen application
and setup that we use for release testing - see:
http://fisheye.jboss.org/browse/Seam/trunk/release-process.txt?r=8379#l274for
the table schema, and information.

741 ms - 1,214 inv. [NAME] java:comp/UserTransaction
20 ms - 1 inv. [NAME] java:comp/env/security/securityMgr
2 ms - 15 inv. [NAME] java:/perfGenEar01EntityManagerFactory

This was about 15 pages with a mix of table navigation, and login/logout,
etc...

-Jay

On Thu, Oct 2, 2008 at 3:50 PM, Jay Balunas <tech4j at gmail.com> wrote:

> Just as a quick test - not my full work up ;-) I ran the dvdstore example
> through some typical work flows to see where and what JNDI calls were made.
>
> This was one user, running the unmodified dvdstore example.
>
> I logged in performed several searches, and made a perchase then logged
> out.  Logged in as manager and reviewed the order and changed the process
> work flow.  Then logged out again.  This  might not be the exact steps, but
> close.
>
> These were the JNDI lookups:
> 971 ms - 2,249 inv. [NAME] java:comp/UserTransaction
> 69 ms - 6 inv. [NAME] jboss-seam-dvd/ShoppingCartBean/local
> 45 ms - 12 inv. [NAME] jboss-seam-dvd/FullTextSearchAction/local
> 5 ms - 1 inv. [NAME] jboss-seam-dvd/CheckoutAction/local
> 3 ms - 6 inv. [NAME] java:comp/env/security/securityMgr
> 2 ms - 17 inv. [NAME] java:/dvdEntityManagerFactory
> 1 ms - 28 inv. [NAME] jboss-seam-dvd/StoreManagerBean/local
> 0 ms - 7 inv. [NAME] jboss-seam-dvd/AuthenticatorAction/local
> 0 ms - 2 inv. [NAME] EJBContext
> 0 ms - 1 inv. [NAME] UserTransaction
> 0 ms - 1 inv. [NAME] env/com.jboss.dvd.seam.StoreManagerBean/em
>
> The "UserTransaction" lookup clearly stands out as a the top offender.
> With multiple users hitting this I think the blocking threads would really
> pile up.
>
> I can run this type of basic test pretty quickly so I am going to do the
> same on the seam-gen project that we use for release testing.  My hunch is
> that the results will be similar.
>
> -Jay
>
>
> On Thu, Oct 2, 2008 at 10:06 AM, Pete Muir <pmuir at redhat.com> wrote:
>
>>
>> On 1 Oct 2008, at 22:53, Jay Balunas wrote:
>>
>>  These finding just cover the wiki application.  We need to do similar
>>> investigations with other applications like dvdstore and chatroom to
>>> exercise some of the other areas.
>>>
>>
>> Yes, as the Wiki is a war, we don't do that many lookups compared to an
>> EJB app.
>>
>
>
>
> --
> blog: http://in.relation.to/Bloggers/Jay
>



-- 
blog: http://in.relation.to/Bloggers/Jay
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.jboss.org/pipermail/seam-dev/attachments/20081002/0a25afcc/attachment.html 


More information about the seam-dev mailing list