0

ZK framework. Performance Test

asked 2006-09-26 20:48:05 +0800

admin gravatar image admin
18691 1 10 130
ZK Team


Orignial message at:
https://sourceforge.net/forum/message.php?msg_id=3933303

By: isobretatel

http://jroller.com/page/galina?entry=zk_framework_performance_test

Conclusions
1. Using the hardware and software, the zkdemo-all.war application can not handle 40 simultaneous active users (CPU overload, max response time is too big).
2. The Page File Usage History shows increasing memory usage, which might indicate memory leaks.

We performed profiling using JBuilder Thread Debugger, which shows that ZK framework has thread contention problems.




delete flag offensive retag edit

7 Replies

Sort by ยป oldest newest

answered 2015-04-20 09:44:07 +0800

Darksu gravatar image Darksu
1991 1 4

Hello cristianrm,

Could you please provide more details on your environment configuration?

Furthermore you could use the following trouble shooting guide to get more information:

http://books.zkoss.org/wiki/ZKDeveloper%27sReference/PerformanceMonitoring/StepbyStepTrouble_Shooting

Best Regards,

Darksu

link publish delete flag offensive edit

answered 2015-04-18 08:59:32 +0800

cristianrm gravatar image cristianrm
12

I'm having serious performance issues, in each session that is connected to the server consumes an absurd amount of memory.

link publish delete flag offensive edit

answered 2006-10-06 20:41:21 +0800

admin gravatar image admin
18691 1 10 130
ZK Team


Orignial message at:
https://sourceforge.net/forum/message.php?msg_id=3950971

By: isobretatel

re-posting email from Henri Chen:

Dear Valery,

We did the performance test ourselves and found an interesting result:

It seems the performance(How many Users) is related to how many requests the Tomcat can handle simultaneously.

We found in your blog an echo2 performance test. It said
echo2 performs to 500 users.
Then we found the echo2 test script sends out only 22 test requests per user while ZK's testing script send out 302 test requests per user.

We did a simple caculation:

1. For echo2: 500 users * 22 requests = 11000

2. For zk: 40 users * 302 request = 12080

3. Your Tomcat server set maxThreads to 10000.

We cut the number of ZK's test requests and try. We found the less test requests sent for a user, the better the Performance.

Here is our 2.1.2 prerelease (
http://zk1.sourceforge.net/zkdemo-all.war ), can you try this on your server and verify whether our observation is correct?

Henri Chen
henrichen <at> potix <dot> com
The ZK Team


link publish delete flag offensive edit

answered 2006-09-29 15:39:04 +0800

admin gravatar image admin
18691 1 10 130
ZK Team


Orignial message at:
https://sourceforge.net/forum/message.php?msg_id=3939361

By: tomyeh

Not a bug. I modified zkdemo not to use BeanShell, extended the size of cache, and some fine tunes. And, the result is great. Henri is preparing a small talk to cover this. Sit tight.

Be aware that this kind of test is somehow misleading (though entertaining for me to tune up for better performance). At least, all frameworks being tested must run in a controlled and equivalent conditions, such as, but not limited to, the same set of functions and operations. Otherwise, it becomes a penalty for being rich -- more features a demo has the worse the performance is.

link publish delete flag offensive edit

answered 2006-09-27 05:58:40 +0800

admin gravatar image admin
18691 1 10 130
ZK Team


Orignial message at:
https://sourceforge.net/forum/message.php?msg_id=3933723

By: tomyeh

This is my first time to hear this. In our own obvervation in the demo site, there are over 70 visitors per hour (200 at peak) and 35 requests per visit are generated per visit. For details, refer to
http://sourceforge.net/forum/message.php?msg_id=3739383

Simon Massey said in another post (
http://sourceforge.net/forum/message.php?msg_id=3911276 ):

"Today I have been aggressively soak testing my refactored AJAX search results page. JMeter has been hammering it for six hours and all is well."


We'll take a look on the test and try to replicate it and see what really happens.
I doubt there is a deep racing bug. Otherwise, it is hard to explain such long maximal response time, while average is still about 0.58 second.

link publish delete flag offensive edit

answered 2006-09-27 05:49:47 +0800

admin gravatar image admin
18691 1 10 130
ZK Team


Orignial message at:
https://sourceforge.net/forum/message.php?msg_id=3933713

By: tomyeh

Defense? No, it is more than welcome for such kind of reports. It is, I believe, why OSS is better, isn't it?


link publish delete flag offensive edit

answered 2006-09-27 03:26:14 +0800

admin gravatar image admin
18691 1 10 130
ZK Team


Orignial message at:
https://sourceforge.net/forum/message.php?msg_id=3933613

By: changhua

Is there any defense from ZK framework?

link publish delete flag offensive edit
Your reply
Please start posting your answer anonymously - your answer will be saved within the current session and published after you log in or create a new account. Please try to give a substantial answer, for discussions, please use comments and please do remember to vote (after you log in)!

[hide preview]

Question tools

Follow

RSS

Stats

Asked: 2006-09-26 20:48:05 +0800

Seen: 761 times

Last updated: Apr 20 '15

Support Options
  • Email Support
  • Training
  • Consulting
  • Outsourcing
Learn More