Current time: 02-21-2018, 09:28 AM Hello There, Guest! (LoginRegister)

Post Reply 
 
Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Test Result Versus Real World Experience
08-11-2016, 01:27 AM
Post: #1
Test Result Versus Real World Experience
I am very grateful to the help that Web Page Test has been over the past several years. We have done extensive work optimizing our web site primarily because we are a software vendor. In looking at my results I see a certain page load value but when I pull that page up in a clean (cache deleted) browser my experience of first view is much faster than reported. My installation is Joomla 3.x running behind NGINX using that cache method. All content is delivered as static files via NGINX and the content is all pushed through port 443 SSL. Here is a most recent test, this time with NGINX set to GZIP which was our final configuration change. OS is Centos 7 64 bit, PHP 5.6, php-fpm, hosted on a dedicated VPS with no other sites by Hostwinds.

Here is a test result, whereas realistic first view comes in at about 3/4 seconds on most of my browsers. Yes, Joomla can be very fast.

Latest test result

Thanks for everyone's feedback.
Find all posts by this user
Quote this message in a reply
08-11-2016, 02:44 PM
Post: #2
RE: Test Result Versus Real World Experience
There's a few variables that can give you different times in the real world:

Location
Are you close to the Dulles WPT location you tested from?
Choose a WPT location closest to your target market to get the most accurate load time experienced by your users, as latency can increase significantly the further the end user is from the server.

Internet Speed
Do you have a faster internet connection?
The WPT default connection speed models a cable 5/1Mbps connection. Increasing this will give you a faster time closer to 3/4s.

CPU power
Testing from a real world desktop will usually render the javascript/css faster than WPT due to having a more powerful CPU.

Other notes:
Be sure to increase the 'Number of Tests to Run' in the advanced settings to run a few tests in case your initial test is an outlier.

As far as improving the current load time you really need to address the 301 redirects (the yellow highlighted rows on the test result). Nearly all the website's resources have a redirect from netdna-ssl.com to your main domain. Ensure they are served correctly from the CDN or remove the CDN.
Visit this user's website Find all posts by this user
Quote this message in a reply
08-12-2016, 03:40 AM
Post: #3
RE: Test Result Versus Real World Experience
Thanks, this makes sense and I will look into those things. Jim

(08-11-2016 02:44 PM)clubberz Wrote:  There's a few variables that can give you different times in the real world:

Location
Are you close to the Dulles WPT location you tested from?
Choose a WPT location closest to your target market to get the most accurate load time experienced by your users, as latency can increase significantly the further the end user is from the server.

Internet Speed
Do you have a faster internet connection?
The WPT default connection speed models a cable 5/1Mbps connection. Increasing this will give you a faster time closer to 3/4s.

CPU power
Testing from a real world desktop will usually render the javascript/css faster than WPT due to having a more powerful CPU.

Other notes:
Be sure to increase the 'Number of Tests to Run' in the advanced settings to run a few tests in case your initial test is an outlier.

As far as improving the current load time you really need to address the 301 redirects (the yellow highlighted rows on the test result). Nearly all the website's resources have a redirect from netdna-ssl.com to your main domain. Ensure they are served correctly from the CDN or remove the CDN.
Find all posts by this user
Quote this message in a reply
08-12-2016, 10:53 PM
Post: #4
RE: Test Result Versus Real World Experience
I reached out to MaxCDN and the issue was that I had set the source URL of the pull zone to plain http. Once I changed it to https the issue went away. The first view time went down by a full second without the 301 redirects. Thanks for pointing me in the right direction! Jim

(08-12-2016 03:40 AM)jimhill10 Wrote:  Thanks, this makes sense and I will look into those things. Jim

(08-11-2016 02:44 PM)clubberz Wrote:  There's a few variables that can give you different times in the real world:

Location
Are you close to the Dulles WPT location you tested from?
Choose a WPT location closest to your target market to get the most accurate load time experienced by your users, as latency can increase significantly the further the end user is from the server.

Internet Speed
Do you have a faster internet connection?
The WPT default connection speed models a cable 5/1Mbps connection. Increasing this will give you a faster time closer to 3/4s.

CPU power
Testing from a real world desktop will usually render the javascript/css faster than WPT due to having a more powerful CPU.

Other notes:
Be sure to increase the 'Number of Tests to Run' in the advanced settings to run a few tests in case your initial test is an outlier.

As far as improving the current load time you really need to address the 301 redirects (the yellow highlighted rows on the test result). Nearly all the website's resources have a redirect from netdna-ssl.com to your main domain. Ensure they are served correctly from the CDN or remove the CDN.
Find all posts by this user
Quote this message in a reply
08-19-2016, 02:14 PM
Post: #5
RE: Test Result Versus Real World Experience
Good to see people using a proper CDN provider!

There's not a huge amount you can do to further improve things really - implement HSTS ( and maybe HPKP if you're feeling brave )... use of http2 lessens the importance of reducing the file count by combining css and js, and the use of sprites where relevant, but I do still like to address it.
Find all posts by this user
Quote this message in a reply
Post Reply 


Forum Jump:


User(s) browsing this thread: 1 Guest(s)