Tricky question

Hi,

This are my test result

1.http://www.webpagetest.org/result/100707_10CD/1/details/

On the first test it is a small gap between line 25 and 26.

2.http://www.webpagetest.org/result/100707_10D1/1/details/

On the 2nd test after I made some optimization ( eliminate header5.jpg -line 9, header-mask-blue.png - line 18) the gap is now between line 20 and 21 and is bigger.

3.http://www.webpagetest.org/result/100707_10FX/1/details/

And after I change the serve of mootools.js from my site to Google Ajax Library the gap is huge almost 1,5 seconds between line 20 and 21 ??

WHY are this gap and why increases as I try to optimize the requests???

Thank you in advance for your answers…

Looks like the CPU spike drops about the same time the gap ends on both test results.

CPU spikes are usually do to JavaScript executing.

You might want to look into your JavaScript to see what is going on behind the scenes.

Yep, the 100% CPU utilization from 2.5 to 4.5 sec probably means you have some heavy (or inefficient) javascript running. It could also be browser layout or other CSS-related things but 9 times out of 10 it is Javascript execution.

The best tool (by far and it’s free) that I have found for helping to track down the hotspot in the js code is Dynatrace Ajax Edition: http://ajax.dynatrace.com/pages/

It is a javascript profiler for IE. I usually just run it against a page, sort the results by JS execution time and drill down into the hotspot. Shouldn’t take more than a few minutes to identify the code at fault.

Thank you Patrick.

I have tryed the Dynatrace but I am not an expert in javascript. from what I see the most delay is on the network , especially on the server.

It is posible that this delay on the server to be generated by javascript or it’s a hosting server problem and I must change the host?

If you are kindley I will be gratefull if you can give me your email to send you the Dynatrace result , maybe from your professional view you can give me some sugestion.

You are my only hope.:s

Feel free to ping me at pmeenan@webpagetest.org (the contact-us link on the site goes to me). I have your page url (unless you are doing a DNS override and testing a dev site) so I can go ahead and run it myself. I have a bunch of meetings coming up but I’ll take a look in a bit.

-Pat

Thank you very much … the investx.net is the live url so you can test when you have time.

Tanks again …

The initialize routine for jaMegaMenuMoo is your code hotspot. Specifically in it’s start routine and this code in particular:

this.items.each (function(li) {

Not sure what functionality that code is providing for your site and what alternatives are available but it is pretty heavy code-wise. If you don’t have any choice but to keep it, maybe initializing it after onLoad would help?

Patrick you have right , that was the problem.

That code is the js of the MegaMenu (the horizontal menu).

I changed the template menu sistem and look what it happend.

http://www.webpagetest.org/result/100709_144Y/1/details/

the problem is gone :smiley:

Tank you Patrick , you are the beeeeessssssssssttttttttt

If you combine some of that JavaScript and CSS it would be even faster.

Congrats on fixing that gap in your code. :slight_smile:

You should also take a look at your back-end. The first byte time is REALLY long and cutting that down will help significantly as it blocks EVERYTHING. Since it’s Joomla you should have a setting to enable a page cache which should get rid of most of the delay (for unauthenticated users anyway).

Tank you guys

I applayed your sugesstions and I gain a lot of speed for my site.

But I don’t undertand something.

I made a lot of test but I receive different results and don’t undarstand why.

With Firefox

1.With www http://www.webpagetest.org/result/100720_1KMX/
2. Withouth www http://www.webpagetest.org/result/100720_1KMV/

With Chrome

  1. With www http://www.webpagetest.org/result/100720_1KMH/
  2. Withouth www http://www.webpagetest.org/result/100720_1KMJ/

Withe IE

  1. With www http://www.webpagetest.org/result/100720_1KMQ/

  2. Withouth www http://www.webpagetest.org/result/100720_1KMK/

  3. The first question is why for tests with www. some or all files are cookiless and for tests withouth www all files have cookies ???:huh:

  4. The second question is why somethimes in the Connection View files are served only from investx.net and other thimes are served from www.investx.net also?? And this it seemes that don’t depend if the test is made with or withouth www. :huh::huh:

As I can see very big differances aren’t between tests, but I also want to know what is the best solution??

  1. Withoth cookies ( as I can see from comparations it seems that test with cookies are sometimes faster )…
  2. With files served only from www.investx.net or also from investx.net. (neither here the tests aren’t clear …in some cases when files are served only from investx.net it seems it is faster …)

I’m not sure what the Firefox, Chrome and IE references are for. WebPagetest doesn’t use your browser for anything but showing you the results. All of the actual testing is done on dedicated machines with IE browsers.

It does look like you have a LOT of variability in the first byte times but that is going to be 100% your back-end platform. Did you look at enabling the joomla caching to see if that helps? I’d also check with the hosting provider to see if the web or (more likely) database server you are on is having performance issues.

-Pat

Ok then the browsers don’t matters, but why the differences that I asked?? Please read again the post now is more complete .

Tanks.

Looks like you started a new thread so I’ll reply over there: http://www.webpagetest.org/forums/showthread.php?tid=311