Hello,
I want to designed a scenario,
Total user are 10 out of that :
| Scenario | Users |
| Login | 100% |
| Search | 80% |
| Create | 20% |
| Logout | 100% |
also explain how to reuse the scenario/script, data correlation between two scenario
How we do it:
1. Create short scenarios (blocks), which logically cover one process: Login, Search, Create, Logout in your case
2. Create global scenarios from blocks, like: Login_Search_Logout, Login_Create_Logout
3. Create load test, which contains required global scenarios:
About data correlation.
We record ONE big scenario, e.g. Login_Search_Create_Logout
Then we make all necessary correlations, data replacing, assign variables etc., and divide this scenario on blocks, like I said above.
Hi,
Thank you a lot for describing the way you are going - this is something that I have interest in.
I will appreciate it if you also can answer these questions:
a) If, say, login sequence changes, do you prefer re-record just it and than make data correlation to match the already existing one made initially or do you prefer to re-record the whole big scenario ?
b) Do you have any processes/enforcing rules/etc. that you follow to guarantee that, say, login sequence recorded a week ago still corresponds to reality? I.e. that requests were not added or removed during past week and that the data sent and received have the same format and contents? Any better practice other than re-record the sequence and compare it with the existing one? Maybe you have something to comment for the https://community.smartbear.com/t5/LoadComplete-Feature-Requests/Make-it-possible-to-compare-and-synchronize-requests-for/idi-p/138803 feature request?
Hi Alex,
First of all I want to clarify, that we perform load testing mostly to control changes outside of API of our applications. E.g. new encryption module was integrated or database version was upgraded or something else. I mean that API itself is rather stable for a very long period. For example authentication calls in our CRM system were not changed for the last 2 years or even more. So I think it easier for us to re-record scenario in the case of significant API changes. Small changes can be fixed manually.
As for your second question. We have a number of API tests, which cover basic functionality and check data structure. We run them before performing load testing to be sure that the application works at all.
Hi,
Are those 20% who do Create execute Search as well? Or do they immediately proceed with Create?
I.e. do you have one sequence like this:
Login(100%) -> Search (100%) -> Create(20%) -> Logout(100%)
or two sequences like this:
Login(80%) -> Search (80%) -> Logout(80%)
Login(20%) -> Create(20%) -> Logout(20%)
?
jawed:
Hi,
I am afraid that the only option for now is to create a test like suggested by baxatob (https://community.smartbear.com/t5/LoadComplete/How-to-configure-mix-scenario-by-reusing-script/m-p/153056#M1629).
I.e. the test that contains two scenarios: first Login_Search_Logout scenario executed by 80 virtual users (VUs) and the second Login_Search_Create_Logout one executed by 20 virtual users. Considering that scenarios within the test are executed in parallel, this should provide you with what you need.
The problem here is that you will need to manually adjust the number of virtual users for every scenario within the test if you decide to change either the total number of VUs or to change the percentage between users who just search and those who search and Create.
If you like, you may create a feature request for this functionality here: https://community.smartbear.com/t5/forums/postpage/board-id/LoadComplete_feature_requests (https://community.smartbear.com/t5/LoadComplete-Feature-Requests/idb-p/LoadComplete_feature_requests/tab/most-recent). (Personally I will vote for it as I consider this to be a great and useful enhancement.)
