process improvment

Desktop Applications Performance Testing Automatization

win311We know, nowadays, how to automatically examine the performance of web applications or simple web pages, it is based mainly on how the web server can cope with a large number of simultaneous sessions/users. There are plenty of professional and open source tools: JMeter, WebLoad, etc., the most of them are targeted to a new technology that is Internet but what about the set of desktop applications: Windows API, Linux (Gnome, KDE) and Mac OS? Small number of tools is the result of little activity on the market or is this just more difficult to implement ? I think that the market will continue to support this type of solutions, the problem is to organize the environment for testing and then automate application runtime.

As we see from the problem can be divided into two parts:

  • Organize test environment
  • Prepare test robot for testing and reporting

If you have done tests with humans (clicking monkeys) we certainly have a infrastructure: ten computers etc, we can now say, “guys go home, do some other interesting work” and then we stay with only the second part of our problem. Of course, someone will ask: “What about the costs? We should give those computers and desks to other employees?” – It can be, and what about the tests? We must start fighting with the help of virtualization, which is not so super cheap and simple and does not allow easy replacement of real machines. We need a very powerful server, or several, software (VM Server is free, but without snapshot feature) and when we have it ready to run, should ask ourselves how closy our environment is to real customer network situations, moreover, how can interferes in this environment which often mutate under many factors.

After a while we have the next requirements:

  • We want to have the real data, or similar to those in which the client works
  • We would like to know what are the measures of success for each performance task
  • We would like to have a multiple test environments: optimal, bad, good, etc.

The problem of a robot that performs a test for the man appears to be something simple: iTestBot , AutoIT and many others. Each of these applications will be able to log the duration of the operation, problems begin when you want to merge the results from many of our machines. Therefore, I recommend to log test results into the database not to a file, a well-designed database of test scenarios and steps will soon achieve a result: for example the average time to open a window with ten thousand records.

Should we forget about desktop applications? Is it worth to invest in testing such solutions? Looking at the Polish market – Yes, we should.

What are Your experiences with desktop testing in the right, automated way 🙂 ?


One comment for “Desktop Applications Performance Testing Automatization”

  1. […] is the original post: Desktop Applications Performacne Testing Automatization Nessun tag per questo […]

    Posted by Mac | Desktop Applications Performacne Testing Automatization | Download Free Software | April 1, 2009, 8:36 pm

Post a comment