|
Fabien Lennertz Posts: 12
7/1/2011
|
Hi Vadim,
Not sure how to put this but I will give it a try...
What is the use of requesting a website with the GET functionality when it only downloads the root?
Lets say we want to load the following website: http://www.stresstimulus.com/
We will use the GET functionality in the [Request Builder] and accordingly press the "Execute" button. The outcome of this will get us 1 primary result. I'm a little confused here. Don't you always want to load the entire page (primary) page including all dependents such as .css and .gif?
On the other hand if you enter the URL in the black bar down below (just above "Capturing"), Fiddler will download the primary page including all the underlying objects. By the way: this can also be realized by enable "Capturing" and enter the URL in your browser. This makes sense. Last question: is there any difference in both options besides the loading the dependent objects?
Regards,
Fab
|
|
|
0
link
|
|
Vadim @StresStimulus Administrator Posts: 583
7/1/2011
|
Hi Fab,
The load testing process includes two distinctive steps:
- Step 1 - Creating a Test Case while navigating by a single user and recording requests.
- Step 2 - Actually running the Test Case under the load (multiple users).
Lets say we want to load the following website: http://www.stresstimulus.com
We will use the GET functionality in the [Request Builder] and accordingly press the "Execute" button. The outcome of this will get us 1 primary result. I'm a little confused here. Don't you always want to load the entire page (primary) page including all dependents such as .css and .gif?
If you decide to use Fiddler’s Request Builder, which can only execute one request at a time, to create a test case (step 1), then you are correct that your test case will be incomplete, since it will not have dependent requests. To fix this problem, don’t use the Request Builder to create a test case.
On the other hand if you enter the URL in the black bar down below (just above "Capturing"), Fiddler will download the primary page including all the underlying objects. By the way: this can also be realized by enable "Capturing" and enter the URL in your browser. This makes sense.
I agree. Navigating in a browser, while capturing in Fiddler all requests, is a better way of creating the test case.
Last question: is there any difference in both options besides the loading the dependent objects?
Both of these options are for the step1, not for the step 2. What matters for the load test is WHAT is in the test case, not HOW it was recorded. In that respect, the Request Builder besides lacking the dependent objects has one additional feature. It allows modifying a single request.
If you need to modify certain requests in the test case, the better option is to record them from the browser, and then to use the technique described in the blog How to edit Test Case requests.
Best,
-Vadim
|
|
|
0
link
|
|
Fabien Lennertz Posts: 12
7/4/2011
|
That makes sense, thanks for your clear response!
Regards,
Fab
|
|
|
0
link
|
|
Fabien Lennertz Posts: 12
7/5/2011
|
Hi Vadim,
Another question. Does it make any difference if a webpage is recorded with different web browsers? Let me put it differently... Can I record a webpage in Internet Explorer and play it back with different browsers (as defined in {StresStimulus]/[Test Confifuration]/[Browser Type] without affecting the reliability of the test results? Or do you recommend: record with IE8 -> playback with IE8, record with Chrome -> playback with Chrome? etc.
Regards,
Fab
|
|
|
0
link
|
|
Vadim @StresStimulus Administrator Posts: 583
7/5/2011
|
Hi Fab,
The objective of the [Browser Type] selection in StresStimulus is testing the impact of different browsers on Web applications from the pure performance standpoint. StresStimulus emulates browsers performance behavior by enforcing HTTP connection limits per host (server) and per-proxy based on a selected web browser. For example, IE9’s connections-per-host limit is 6 and connections-per-proxy limit is 12, while Firefox’s 3 is 6 and 8 respectively. In that respect, the answer to your question is that, to adequately evaluate many browsers’ performance impact, you can record a webpage in Internet Explorer and play it back with these browsers.
Another aspect of the browser’s impact on a web application is the user agent header. Some of the web applications are sensitive to the user agent, and arbitrary changing it can break the application. Therefore the current version of StresStimulus does not replace the recorded user agent to avoid such incidents. Because some of the web apps render different pages based on a user agent, the coming version of StresStimulus will modify the user agent according to the selected browser type, but it also will give a safeguard option to keep the recorded user agent unchanged.
Update: The option to replace the user-agent string is now configurable.
Best,
-Vadim
|
|
|
0
link
|