CAS Performance

Wed Nov 15 10:31:58 EST 2006

I am going to change my custom ticket registry so that it uses a HashMap again. Can I just drop the synch on the method or do I need to move it to a synch on the ticket?

Thanks,
Pete

-----Original Message-----
From: cas-bounces at tp.its.yale.edu [mailto:cas-bounces at tp.its.yale.edu]On Behalf Of Scott Battaglia
Sent: Wednesday, November 15, 2006 10:13 AM
To: Yale CAS mailing list
Subject: Re: Re: Re: Re: CAS Performance

I just realized that the default registry had synchronized methods so I put in a JIRA issue to remove them (I didn't know they were there as I hadn't added them and after discussion with another coworker where we traced through we concluded they are not needed in that implementation).  

Also, I'm not sure how fast JCS is.  But CAS itself is pretty lightweight and we're synchronizing on individual tickets so unless all of your load testing is sharing one TGT that shouldn't affect anything.

-Scott

On 11/14/06, Giesin, Peter (NBC Universal) < Peter.Giesin at nbcuni.com> wrote: 

I am using 3.0.5. Yes, that looks true. I just can't seem to get any performance out of this thing. Have you done any type of load testing? What type of performance have you seen? If I can't get this this work any better I an going to have to pull it and replace it with something else. I just can't seem to get more than 20-30 login requests per second.

Pete

-----Original Message-----
From: cas-bounces at tp.its.yale.edu  <mailto:cas-bounces at tp.its.yale.edu> [mailto: cas-bounces at tp.its.yale.edu]On Behalf Of Scott Battaglia
Sent: Monday, November 13, 2006 11:08 PM
To: Yale CAS mailing list
Subject: Re: Re: Re: Re: CAS Performance

Which version of CAS are you using?  Later versions synchronize on the specific ticket and not on any registries.

-Scott

On 11/13/06, Pete Giesin < pgiesin at hubcitymedia.com> wrote: 

After looking at it some more I think the log entries are not related.
One is the login and the other is the logout that is being processed.

The concern that I have is the large number of synchronized blocks
that are being used. This seems like it will cause many bottlenecks. I
updated the DefaultTicketCache to use JCS and removed the synchronized
methods. This seemed to help out some, but not entirely.

Thanks,
Pete

On 11/13/06, Scott Battaglia < scott.battaglia at gmail.com> wrote:
> Pete,
>
> Does that happen all the time?  (and is this only with one user?).  The 
> result of an "authenticatedButNoService" call is a request to render the
> viewGenericLoginSuccess page. Rendering a genericLoginSuccessView should not
> result in a ticket being removed.  Tickets are only removed if you do a 
> renew=true with different credentials or if logout is called (unless I am
> remembering wrong).
>
> Also, CAS itself is not very memory intensive so 4 GB of memory should be
> more than sufficient.  We did find in our testing (which we did on a 
> single-CPU machine) that the majority of the CPU time was being taken up by
> Apache processes (but then we had an underpowered machine).
>
> -Scott
>
>
> On 11/13/06, Pete Giesin < pgiesin at hubcitymedia.com> wrote:
> >
> > More specs: dual AMD dual core cpu's with 4GB of memory.
> >
> > One interesting thing is the 2 second delay between these  actions: 
> >
> > 2006-11-13 17:02:27,856 DEBUG
> > [org.jasig.cas.web.flow.HasServiceCheckAction] - <Action
> > 'HasServiceCheckAction' completed execution; result is
> > 'authenticatedButNoService'> 
> > 2006-11-13 17:02:29,792 DEBUG
> > [org.jasig.cas.CentralAuthenticationServiceImpl] -
> <Removing ticket
> > [TGT-369-p62ifeG755sWev2LdE1Cl9pjPmUFJKeZMzk-50] from
> registry.> 
> >
> > Pete
> >
> > On 11/13/06, Scott Battaglia < scott.battaglia at gmail.com> wrote:
> > > Do you hav any more information on the specs? 
> > >
> > >
> > > On 11/13/06, Pete Giesin < pgiesin at hubcitymedia.com> wrote:
> > > > HP blade running Redhat Linux for 64-bit. 
> > > >
> > > > On 11/13/06, Scott Battaglia < scott.battaglia at gmail.com> wrote:
> > > > > What type of server is it?
> > > > >
> > > > > -Scott
> > > > >
> > > > >
> > > > > On 11/13/06, Pete Giesin < pgiesin at hubcitymedia.com  <mailto:pgiesin at hubcitymedia.com> > wrote:
> > > > > >
> > > > > > I am performing some load testing against CAS and am getting
> terrible
> > > > > > results. Something like 30 requests per second. I have CAS running 
> on
> > > > > > a single server using Apache + Tomcat. I have a JDBC datasource
> > > > > > defined for the user lookups. I also have an ehcache defined to
> cache
> > > > > > the users. I was wondering if anyone has any suggestions on how to
> > > > > > tweak the configuration in order to increase the performance. My
> > > > > > objective is 100 requests per second. 
> > > > > >
> > > > > > Thanks,
> > > > > > Pete
> > > > > > _______________________________________________
> > > > > > Yale CAS mailing list 
> > > > > > cas at tp.its.yale.edu
> > > > > > http://tp.its.yale.edu/mailman/listinfo/cas  <http://tp.its.yale.edu/mailman/listinfo/cas> 
> > > > > >
> > > > >
> > > > >
> > > > > _______________________________________________
> > > > > Yale CAS mailing list
> > > > > cas at tp.its.yale.edu
> > > > > http://tp.its.yale.edu/mailman/listinfo/cas
> > > > > 
> > > > >
> > > > >
> > > > _______________________________________________
> > > > Yale CAS mailing list
> > > > cas at tp.its.yale.edu
> > > > http://tp.its.yale.edu/mailman/listinfo/cas
> > > >
> > >
> > >
> > > _______________________________________________ 
> > > Yale CAS mailing list
> > > cas at tp.its.yale.edu
> > > http://tp.its.yale.edu/mailman/listinfo/cas  <http://tp.its.yale.edu/mailman/listinfo/cas> 
> > >
> > >
> > >
> > _______________________________________________
> > Yale CAS mailing list
> > cas at tp.its.yale.edu  <mailto:cas at tp.its.yale.edu> 
> > http://tp.its.yale.edu/mailman/listinfo/cas
> >
>
>
> _______________________________________________
> Yale CAS mailing list 
> cas at tp.its.yale.edu
> http://tp.its.yale.edu/mailman/listinfo/cas
>
>
>
_______________________________________________ 
Yale CAS mailing list
cas at tp.its.yale.edu
http://tp.its.yale.edu/mailman/listinfo/cas

_______________________________________________
Yale CAS mailing list
cas at tp.its.yale.edu
http://tp.its.yale.edu/mailman/listinfo/cas

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://tp.its.yale.edu/pipermail/cas/attachments/20061115/92dfba73/attachment.html