see my answers below.
Question:
A Web page in Jahia can contain one or several images or other files (css, js, etc). Is this true to say that a page could take one minute to load for a user if this page contains 5 images?
A Gaussian random timer was added to the thread with a deviation of 10 sec (10'000ms) and a constant delay offset of 10 sec, so a "request" is made every 0 to 20 seconds, but unless mistaken that does not apply to embedded resources. I think this make more sense as you want to simulate real usage, and as a real user you will request all resource on the same page at approximately the same time.
Example: A page contains 5 images. So in Jmeter this is 6 HTTP requests.
Page 1 : Random time from 0 to 20 seconds.
Image 1: Random time from 0 to 20 seconds.
Image 2: Random time from 0 to 20 seconds.
Image 3: Random time from 0 to 20 seconds.
Image 4: Random time from 0 to 20 seconds.
Image 5: Random time from 0 to 20 seconds.
We could say that this example page can take between 0 second and 120 seconds to be completely obtain with an average of 60 seconds.
As mentionned above, a call to a page is made every 0 to 20 seconds, embedded resources are not subject to the 20 seconds delay.
How did you obtain your statistics by page if in JMeter we refer to http requests and that a page can consist of several http requests? Did you omit the images and the other files of the Web pages?
I'm not sure if the option to retrieve the image was flagged or not, but as the test was performed on average pages with not much images (except one page), I'm not sure this would have made a large amount of difference.
Anyway the benchmark is a year old and was performed on a 15 months old release, a new benchmark should be performed on a more up2date release to be more meaningful.
Michel
