tomcat-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Peter Lin <tcw00l...@yahoo.com>
Subject RE: dramatic performance differences on development machines
Date Tue, 13 Jan 2004 16:12:26 GMT
it would really be helpful if you clearly state the systems and the performance. information
like
 
1. CPU
2. ram
3. HD
4. OS version
5. tomcat version
6. jdk version
7. how you're measuring the performance
8. type of ethernet
9. kind of router
10. free memory before tomcat starts
11. peak memory usage under load testing
12. list of other processes
13. configuration differences between the systems
14. the performance of each machine
15. rough description of what the page does in functional terms. not enough to get you in
trouble, but enough for others to compare it to similar types of webapps.
 
at this point, it's like trying to hit a bullseye from 5K feet in the dark with a 22 calibre
pistol.
 
peter lin


Johan Coens <j.coens@thefactore.com> wrote:
Unfortunattaly no option, we are bound to a specific java version and tomcat
version to gain support from the supplier...

-----Original Message-----
From: Edson Alves Pereira [mailto:eapereira@panamericano.com.br]
Sent: 13 January 2004 17:51
To: 'Tomcat Users List'
Subject: RE: dramatic performance differences on development machines


Try tomcat-4.1.29 or tomcat-5.x

> ----------
> De: Johan Coens[SMTP:j.coens@thefactore.com]
> Responder: Tomcat Users List
> Enviada: terça-feira, 13 de janeiro de 2004 11:10
> Para: Tomcat Users List
> Assunto: RE: dramatic performance differences on development machines
>
> A lot of network traffic, database lookups, rmi is going on, but i even
> tested that by placing code on the mediasurface server with oracle
> (placing
> all on one box), and so limiting network traffic, but same performance
> issues occured.
>
> I tested on websphere and everything is speedy (so, its not the code).
>
> It should be something in the system, a configuration which influences
> tomcat performance dramatically and not webspheres performance, but what i
> can't figure out what this could be. Maybe the only solution for me is
> running websphere...
>
> Thanks for the feedback,
> Johan
>
> -----Original Message-----
> From: Donie Kelly [mailto:donie.kelly@tecnomen.ie]
> Sent: 13 January 2004 14:48
> To: 'Tomcat Users List'
> Subject: RE: dramatic performance differences on development machines
>
>
> What does the servlet do during the request. Is it a database lookup or
> what? It's intresting that it's using 50-90% cpu time. If it were a
> network
> error or mis-configuration I'd expect to see 0% cpu used during the
> timeout
> period. Are you sure the application is working correctly when the
> response
> has come back. Maybe there is some sort of timeout running in your
> application that does not yield very well, ie: a tight loop waiting for
> something? Maybe your processing is not as correct as you think.
>
> Give us more to work with...
> Donie
>
> -----Original Message-----
> From: Ralph Einfeldt [mailto:ralph.einfeldt@uptime-isc.de]
> Sent: 13 January 2004 13:45
> To: Tomcat Users List
> Subject: RE: dramatic performance differences on development machines
>
> This kind of performance degration can
> also have some of the following causes if
> the load of the system doesn't indicate
> a problem:
>
> - long or failing DNS Lookups.
> - Missconfiguration that leads to round trips
> in the network.
> - locks (e.g. Database)
>
> I think you have to isolate one request that
> takes long and find out where the time is spent.
> (This doesn't mean in all cases profiling, in the
> first step it might be enough to find out if the
> time is spent before the request reaches the
> application, in the application, or after the
> application has sent the response.)
>
> > -----Original Message-----
> > From: Johan Coens [mailto:j.coens@thefactore.com]
> > Sent: Tuesday, January 13, 2004 2:28 PM
> > To: Tomcat Users List
> > Subject: RE: dramatic performance differences on development machines
> >
> >
> > Here the specs are:
> >
> > It's a windows XP development Client
> > Pentium 4, 2GHz, 512Mb memory
> > jdk 1.3.1_06
> > tomcat 4.0.6
> >
> > Tomcat is consuming 50-90% of processing time when serving
> > the request.
> > Notice, i tetsted the app on websphere too, it is serving
> > quite fast, 400ms.
> > instead of 20000ms.
> >
> > If anybody can point me where too look at I would be very happy.
> >
> > Johan
> >
> > -----Original Message-----
> > From: Donie Kelly [mailto:donie.kelly@tecnomen.ie]
> > Sent: 13 January 2004 12:55
> > To: 'Tomcat Users List'
> > Subject: RE: dramatic performance differences on development machines
> >
> >
> > Any chance there is something else running on the machine
> > that's killing the
> > performance.
> >
> > You should post the specs of the machine if you expect a
> > reasonable guess as
> > to your problem.
> >
> > Donie
> >
> > -----Original Message-----
> > From: Nikola Milutinovic [mailto:Nikola.Milutinovic@ev.co.yu]
> > Sent: 13 January 2004 09:43
> > To: Tomcat Users List
> > Subject: Re: dramatic performance differences on development machines
> >
> > Johan Coens wrote:
> >
> > > Hello Nikola,
> > >
> > > Machines are not identical, the fast machine has different
> > specs (less
> > > memory, less disk space and less cpu) then the slow machine
> > (this one has
> > > better specs).
> >
> > Quite ironic.
> >
> > > One hint we've got is the carachter encoding in which the
> > > file is saved, but it seems to me this cannot be the problem...
> >
> > It can be a problem, but not responsible for 20x degradation.
> >
> > > Sure, heavy
> > > artillery can be used, but i don't think that would lead us
> > to a solution,
> > > also because the machine with lesser specs serves better,
> > and we use the
> > > same tomcat version, same settings and same jdk version.
> >
> > Agreed. The only thing you're left with is profiling. There
> > were some posts
> > on
> > that subject. So far, we've heard of JProfiler and something from IBM.
> > Borland's
> > JBuilder has OptimizeIt Suite", but it costs $$$.
> >
> > Nix.
> >
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: tomcat-user-unsubscribe@jakarta.apache.org
> > For additional commands, e-mail: tomcat-user-help@jakarta.apache.org
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: tomcat-user-unsubscribe@jakarta.apache.org
> > For additional commands, e-mail: tomcat-user-help@jakarta.apache.org
> >
> >
> >
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: tomcat-user-unsubscribe@jakarta.apache.org
> > For additional commands, e-mail: tomcat-user-help@jakarta.apache.org
> >
> >
> >
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: tomcat-user-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: tomcat-user-help@jakarta.apache.org
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: tomcat-user-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: tomcat-user-help@jakarta.apache.org
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: tomcat-user-unsubscribe@jakarta.apache.org
> For additional commands, e-mail: tomcat-user-help@jakarta.apache.org
>



---------------------------------------------------------------------
To unsubscribe, e-mail: tomcat-user-unsubscribe@jakarta.apache.org
For additional commands, e-mail: tomcat-user-help@jakarta.apache.org


---------------------------------
Do you Yahoo!?
Protect your identity with Yahoo! Mail AddressGuard
Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message