Performance Testing Using TC9 or 10

Hi Azeddin,



> Can I use Testcomplete 9 or 10 for performance testing?.

I would like to ask you to explain what you need with more details.

You can use TestComplete to measure performance from the end-user point of view. E.g. you can measure how much time does it take from the moment when user pressed the Save button and until the confirmation message is displayed on the screen.

Using the Network Testing TestComplete's functionality (http://support.smartbear.com/viewarticle/62722/ and http://support.smartbear.com/viewarticle/55046/) you can measure the same when a limited number of users (2-5, depending on the number of TestComplete licenses you have) do the same action concurrently.

If you need to measure the performance of the code to find, for example, the slowest function, than you need to use AQtime, profiler by SmartBear (http://smartbear.com/products/qa-tools/application-performance-profiling/).

If your tested application is web-based and you need to test the performance of the server and how quickly it responses to the application's requests, then you may consider LoadUIWeb (to check from the end-user point of view, http://smartbear.com/products/qa-tools/load-testing/) or SoapUI / LoadUI pair (http://smartbear.com/products/qa-tools/web-service-testing-tool, http://smartbear.com/products/qa-tools/web-service-load-testing) to do testing on the API level.

You can get some basic performance metrics. Mainly at the client end.



There are also some performance counters you can switch on during the test run. Again, mainly to monitor things like memory usage, CPU usage etc, at the client end. Although, this is all stuff you can also do with built in windows tools like perfmon.



For proper, scalable, load/performance/stress testing, TestComplete is not the right tool.

I created some UI-based performance testing for our apps with TC9 and TC10 and have been very successful.

Which is fine at client end.



But if you're only running as a single user, you need to simulate real world conditions. Which will more often than not involve multiple users running simultaneously.



For a standalone single user application I can see the validity.



For a client/server based system with a scalable user base, I generally wouldn't be using TestComplete.



So it really depends what you're trying to do. Which the OP doesn't say.



"Performance testing" is far too general a term to get an accurate answer ....