I'm a newby on QARun (4.8) and am wondering how other people use the product to develop testing scripts.
The environment: We have a development team programming a multi-tier java client-server application. The basic idea is to do some performance testing on 100 robotic clients spread all over the country. Due to the mostly very slow network connections right now we plan to do the testing script development on one client and roll out the resulting access-mdb to all 100 clients.
Every programmer will be responsible that as soon as new components of the program are available, testing scripts will also be provided. This means that many people will somehow have to work on the same MDB.
How do other people implement some kind of "release management" for the testing scripts? I was thinking of every programmer having a unique basic-mdb, containing generic errorhandling, logon and logoff-procedures, ... An additional programmers-mdb would contain the individual testing scripts. Is that realistic? How could all the individual mdbs be put together?
Is there any chance to tell runawl to use a different mdb?
Best thanks in advance for your ideas and your experience...Andreas
had to switch identity for a moment (forgot my password and don't have access to the email-account connected to "kuffi")
what I learned from reading:
- runawls database is only changable through regedit (is that still true for 4.8?)
- some people are using more than one "default" database
- Release management is often done with an selfwritten external database, no useful tools provided by QARun.
Still I'd like to know what in experienced QARunners opinion would be the best way of setting up QARun ecspecially in our environment (slow network connection, several people writing performance testing scripts)
Do you realy want to do performance tests with QARun?
In my point of view, QARun is for functional testing. I used it as well for performance test, but i would prefer QALoad or LoadRunner for that. Those Tools are based on the Network Protocol and you only need 1 client to simulate your 100 users.
Additional, you can work together with network agents. that means you can measure on each tier your network traffic and performance of your application (DB-Server, Application-Server etc.).
If you really wont to use QARun for that, do that what QAGuy allready mentioned. But if you use a central database and having your clients in the whole country, do not forget the network latency to check out the scripts and writing the logs. i had that one time, and it tooks me arround 1 minute to load the script from the db-server into qarun!
[This message has been edited by Adrian Z (edited 11-21-2002).]
Thanks for the advise so far.
In our case performance testing is done by the application itself. It is ARM-instrumented and writes all performance related data to a central database (using Tivoli TAPM). We need QARun to simulate a human user working with the application. The reason for doing this with 100 roboters is that we need the performance data for every single departement. This is what I meant with performance testing.