Two Response Time Question

[Guest twilliams]

I have two questions regarding the response times I'm seeing on a recent load test. First, a bit of background info on the test: Test configured to run until 300 users complete. Only one profile used consisting of (1) browsing to a main website page, then (2) simply clicking a category search button. Here are my questions.

 

1. Does this "run until all users complete" launch all users immediately? Or does it stagger the number of sessions? The results I'm seeing seem to show all users running at once with a bunch timing out and then retrying, is this accurate?

 

2. More importantly, this one is throwing me off. Using just the first (main) page of this test as an example, I see zeros in the report for the time frames 00:00-00:55 and through 00:55-01:50. What this tells me is that the server did not respond with any data during these times, right? Then in the 01:50-02:45 time frame I see an average response time of 78.4 seconds... How could the response time have been 78 seconds if it's within the 1:50 to 2:45 time frame? Am I reading this wrong?

 

Thanks in advance! I really like this product and understanding these two essential pieces of information will really help us get the most for our money.

 

Trent

[sergei 0]

1. Does this "run until all users complete" launch all users immediately? Or does it stagger the number of sessions? The results I'm seeing seem to show all users running at once with a bunch timing out and then retrying, is this accurate?

 

"Complete all open sessions" option means that after time of test is over all opened sessions should be executed to the end. A number of started users is regulated by "User Load" options.

 

2. More importantly, this one is throwing me off. Using just the first (main) page of this test as an example, I see zeros in the report for the time frames 00:00-00:55 and through 00:55-01:50. What this tells me is that the server did not respond with any data during these times, right? Then in the 01:50-02:45 time frame I see an average response time of 78.4 seconds... How could the response time have been 78 seconds if it's within the 1:50 to 2:45 time frame? Am I reading this wrong?

 

Average time is available only after whole request with all page elements was completed. Therefore you see results only in 1:50 - 2:45 time frame.

 

[Guest twilliams]

"Complete all open sessions" option means that after time of test is over all opened sessions should be executed to the end. A number of started users is regulated by "User Load" options.

 

Average time is available only after whole request with all page elements was completed. Therefore you see results only in 1:50 - 2:45 time frame.

 

So I set my user load option to 300, then it'll start all 300 immediately and run until it finishes?

 

"after whole request with all page elements was completed" - This makes perfect sense, I'm still a little confused though, if the response time was 78 seconds and it appears in the 1:50 column, what was it doing for the first 32 seconds?

 

Thanks for the quick reply!

[sergei 0]

So I set my user load option to 300, then it'll start all 300 immediately and run until it finishes?

 

Yes. All 300 virtual users will be started simultaneously. Each virtual user during test time will start a new session after completing the previous one.

 

"after whole request with all page elements was completed" - This makes perfect sense, I'm still a little confused though, if the response time was 78 seconds and it appears in the 1:50 column, what was it doing for the first 32 seconds?

 

Thanks for the quick reply!

 

First 78 seconds WAPT requested and downloaded main HTML code of the page and rest of the time it downloaded page elements: pictures, css and scripts.

[Guest twilliams]

Yes. All 300 virtual users will be started simultaneously. Each virtual user during test time will start a new session after completing the previous one.

 

This sounds contradictory to the description in the program. You're saying 300 users will start immediately and as one finishes the test case it'll run it again? If that's the case, how does it know to finish? When I'm configuring the scenario there is no place to configure a test run time, there is only an option for "run until user sessions complete" and a number of users.

 

First 78 seconds WAPT requested and downloaded main HTML code of the page and rest of the time it downloaded page elements: pictures, css and scripts.

 

This is still very confusing to me. Sorry if I'm slow and not getting the point . If I see a 78 second response time in the 1:50 - 2:55 column, am I to assume it was truly a 78 second response time or can I assume it actually took at least 1:50 seconds to serve up all page content? If the former, then why doesn't it appear in the 1:00 - 1:50 column?

 

Thanks again for your help, I really want to understand this better.

[sergei 0]

This sounds contradictory to the description in the program. You're saying 300 users will start immediately and as one finishes the test case it'll run it again? If that's the case, how does it know to finish? When I'm configuring the scenario there is no place to configure a test run time, there is only an option for "run until user sessions complete" and a number of users.

 

In this case the test will be finished when the required number of sessions will be completed. For example, you configured 300 users to run and 500 sessions to complete. 300 users will start the first session and only 200 will start the second session after complition the first, then the test will be finished.

 

If I see a 78 second response time in the 1:50 - 2:55 column, am I to assume it was truly a 78 second response time or can I assume it actually took at least 1:50 seconds to serve up all page content?

 

You can consider it as 78 seconds user waits to see a HTML page without pictures and at least 1:50 to see all pictures on the page.

 

I see there are some possibilities for improvment in the future versions.