I would also read about "how to scale up web applications" !
Also you may talk to those which have their web apps already in the cloud ! 
In a cloud, adding more CPU's, adding more memory, adding more data storage 
space is easy.
And what about communication and data band width and related equipment ? 
any clues about how large or small your request/reply packages are ?
It is a difference if 5000 users download a streaming movie in real-time, or 
streaming music in real-time, or streaming compressed stock market data in even 
better real time or books or any other sort of compound documents in 8 to 10 
minutes down load time,; so what are the demands from a user perspective to 
reach a high quality of service?
Josef Stadelmann

-----Ursprüngliche Nachricht-----
Von: Mark Eggers [mailto:its_toas...@yahoo.com] 
Gesendet: Mittwoch, 31. Juli 2013 00:18
An: Tomcat Users List
Betreff: Re: Configuration question for 2500 simultaneous users.

On 7/30/2013 1:17 PM, Tomcat Random wrote:
> Thanks Mark, I will give it a close read.
>
> As far as profiling, are you using any tools that are worth mentioning?
>

Nothing outstanding, since currently all of our applications are pretty 
lightweight. That may change if we redo the architecture.

JMeter / Selenium in combination can generate a lot of traffic. Generate a 
selenium test script, export to JUnit, couple with HTMLUnit, and hammer away.

There are several ways to watch what goes on with your application:

JConsole
VisualVM

The Tomcat Wiki page has more:

http://wiki.apache.org/tomcat/FAQ/Monitoring

For lighter weight profiling (usually to figure out where the application 
bottlenecks are), I run the project under NetBeans and instrument the project.

Access logs are usually a good first source for generating JMeter tests.

In general, people can only give you guidelines concerning sizing, profiling, 
and benchmarking. The particulars depend on your particular application.

. . . . just my two cents.
/mde/

PS - Please don't top post.

> Best,
> A
>
>
>
> On Tue, Jul 30, 2013 at 4:02 PM, Mark Eggers <its_toas...@yahoo.com> wrote:
>
>> On 7/30/2013 12:42 PM, Tomcat Random wrote:
>>
>>> The project I'm working on has 5000 simultaneous users average. I 
>>> have two physical servers both running an instance of Tomcat 7.0. 
>>> They're behind a physical load balancer with sticky, least 
>>> connections balancing. Nothing in front of the Tomcats. Port 80 to 
>>> is routed to them by iptables.
>>>
>>> Anyone out there willing to offer some tips (or point me to them) on 
>>> configuration for this amount of traffic?
>>>
>>> Environment is:
>>> DELL PowerEdge R720 - 32 GB DELL RAM, GB Memory: 32 Single Socket 
>>> Six Core Intel Xeon E5-2640 2.5GHz, #Processors: 1, #Cores per Proc: 
>>> 6 RHEL 6
>>>
>>> TIA,
>>> Alec
>>>
>>>
>> A great overview, and a solid outline of the process you should follow:
>>
>> http://people.apache.org/~**markt/presentations/2009-04-**
>> 01-TomcatTuning.pdf<http://people.apache.org/~markt/presentations/200
>> 9-04-01-TomcatTuning.pdf>
>>
>> That, plus profiling your application with real-world traffic to 
>> understand bottlenecks and use cases . . .
>>
>> . . . just my two cents.
>> /mde/


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
For additional commands, e-mail: users-h...@tomcat.apache.org


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
For additional commands, e-mail: users-h...@tomcat.apache.org

Reply via email to