1
avtx30
About Xoops(es) perfomance
  • 2008/4/14 23:00

  • avtx30

  • Not too shy to talk

  • Posts: 181

  • Since: 2006/10/12


I've been playing with XOOPS for 3 years and love 'them' very much. I said 'them' because there are alot of XOOPS out there which some of you may not notice.

I started with XOOPS 2.0.13.2 then stayed until XOOPS 2.0.17 before switching to ICMS which gives me better and nicer Admin Control Panel (ACP) as well as some great commitments - one of them is to stay tuned for GIJOE's modules. GIJOE has also stated that he will support ICMS.

Lately I started to play with XOOPS Cube Legacy (XCL) and it turns out that XOOPS Cube wins others of its sophisticated technologies such as Class Override, Preload, and Delegate. I think XOOPS Cube is a very good example application for developers who can hack the core just by writing 'preload', without touching a single line of the core's codes. On the other hand XOOPS Cube seems to be slow that can be noticeable. This is the only disavantage of XOOPS Cube which hopefully will be cleared in the near future.

Beside of Xoops, ICMS, and XCL, I also play with Xoops2.0.16-JP, XOOPS Mexico, and Simple-Xoops. No specific reasons especially any 'political' reasons, just for fun. Just because I love XOOPS very much...

There are some works recently made XOOPS faster. This made me to wonder which XOOPS is the fastest without taking into account their functional capabilities. And I did a test yesterday to answer the question myself. And the conclusion is while XOOPS 2.0.18.1 gives good performance, Xoops2.0.16a-JP is the fastest and XOOPS Cube as expected the slowest.

Environments
- XOOPS Server?Linux x32 on a box with dual-core CPUs and 1GB of RAM, Apache 2.2.8, MySQL 5.0.58, PHP 5.2.5
- Test client: Windows XP SP2
- Test tool: JMeter 2.3.1

Server and test client are in the same local network without access from external.

Xoops Cores
- 2.0.16a-JP: Based on XOOPS 2.0.13 with extensive enhancements by Japanese XOOPS team
- XOOPS Cube 2.1.4: The combination between Cube core and Legacy module that emulates the Xoops's 'system' module. XOOPS Cube is written from scratch by some greatest XOOPS developers (minahito, onokazu, GIJOE,...)
- 2.0.18.1: Xoops.org latest stable version. The resource.db.php bug fixed.
- 2.2.6: Xoops.org 2.2 series latest version
- 2.3.0-alpha2: Xoops.org latest development version
- XOOPS Mexico: Works by some developers in Mexico based on XOOPS 2.0.16. The resource.db.php bug fixed.
- Simple-Xoops 0.8.2: Works by some developers in Germany based on XOOPS 2.2.5
- I.mpressCMS 1.0: The lastest fork from XOOPS 2.0.17 with alot of enhancements

Xoops Modules
I installed following modules to every cores. Test data is batch input via phpmyadmin.

?Protector 3.16
?Altsys 0.57
?Xigg 1.0.0b1
?XSNS 1.1.0
?D3downloads 0.60d
?mylinks 1.1.0

Test method
I issued HTTP requests (method: GET) targeting following URLs then recorded the average response time (in miliseconds). Number of threads: 50, Ramp-up Period?300 seconds, Loop Count: 2.

{XOOPS_URL}/index.php
{XOOPS_URL}/modules/Xigg/index.php/node/1
{XOOPS_URL}modules/d3downloads/index.php
{XOOPS_URL}/modules/mylinks/index.php

Test results
Here are the results. The shorter response time, the faster Xoops. You will see Xoops2.0.16a-JP is the fastest then XOOPS Mexico, XOOPS 2.0.18.1,...and XOOPS Cube 2.1.4 is the slowest.

1) XOOPS 2.0.16a-JP: 279 ms
2) XOOPS Mexico: 279 ms
3) XOOPS 2.0.18.1: 296 ms
4) XOOPS 2.3.0-apha2: 307 ms
5) I.mpressCMS 1.0 Final: 321 ms
6) XOOPS 2.2.6: 326 ms
7) Simple-Xoops: 0.8.2 349 ms
8) XoopsCubeLegacy 2.1.4: 397 ms

Others
Sorry for my poor English.

Feedbacks are welcome!

Demos sites here: http://www.nhatban.net/nb/content/index.php?content_id=21

Thanks for reading.

2
giba
Re: About Xoops(es) perfomance
  • 2008/4/15 4:18

  • giba

  • Just can't stay away

  • Posts: 638

  • Since: 2003/4/26


ok,

I think that failed 2 work in this test, though it would not change the result in speed if compared to the XTXM xoopers here:

http://www.xoops.net.br/

But this other original work done by wizanda. An original distribution of 2.2 but custom and prioritizing speed. I have tested and assure that it is about 4 times faster on the basic functions of work.

http://wizanda.svn.sourceforge.net/viewvc/wizanda/finalize/


If he can perform the test, the result is welcome.


You can check in the roadmap and also in our wiki.

Ps: This version XTXM xoopers was placed the provision of the entire team to develop the core to analyze what would be good to include in the core.

In any case, there remains a willingness to use what is necessary.

ps2: Because of censorship exists here today, withdrew a few sentences to avoid any kind of change in my answer.

3
avtx30
Re: About Xoops(es) perfomance
  • 2008/4/15 9:27

  • avtx30

  • Not too shy to talk

  • Posts: 181

  • Since: 2006/10/12


@Giba

Thanks for the comments and pointing out the works by wizanda. A moderator here also told me that existed a fork called XTM. I will look into both of them later.

I post the test result here hoping people will have something to consider when choosing a XOOPS core. Functions are very important but speed is sometime more important.

In the future I will report test results about the effects of other technics to make XOOPS faster (Xcache, PHP 5.3, php_speedy, XOOPS templates cache, MySQL tuning, etc.)

PS. This is the first time I used JMeter. I found it's very powerful test tool. Why don't you give it a try to test things yourself?

4
giba
Re: About Xoops(es) perfomance
  • 2008/5/7 10:55

  • giba

  • Just can't stay away

  • Posts: 638

  • Since: 2003/4/26


It will not include this work in the tests?

Quote:




http://www.xoops.net.br/

But this other original work done by wizanda. An original distribution of 2.2 but custom and prioritizing speed. I have tested and assure that it is about 4 times faster on the basic functions of work.

http://wizanda.svn.sourceforge.net/viewvc/wizanda/finalize/


Login

Username:
Password:

Lost Password? Register now!

Who's Online

70 user(s) are online (38 user(s) are browsing Support Forums)


Members: 0


Guests: 70


more...

Donat-O-Meter

Stats
Goal: $100.00
Due Date: May 31
Gross Amount: $0.00
Net Balance: $0.00
Left to go: $100.00
Make donations with PayPal!

Latest GitHub Commits