Is load runner used to automate the web application, or only used to performance .

Question

I have tried the load runner for performance tool till now, my friend suggest that load runner will also used for automate the web application like QTP ,selenium . any one please share your knowledge in the above criteria .

Answer 1

Your friend is architecturally challenged. Look to the OSI model and your client architecture. LoadRunner, for the majority of the protocols, does not operate at the top of the OSI application layer, where functional automation tools operate.

With a few exceptions for specific needs and interfaces, performance testing tools as a class are designed to operate at a lower layer of the OSI model than the top of the application layer. They are designed to reproduce the messages that an application sends to a server, not necessarily to reproduce the clicks of a mouse on a dialogue box (unless that is the only required method to generate a message and load). GUI's tend to be very expensive, and in some cases you cannot even run more than one instance of a GUI per host, so having the ability to strip out the generation of messages from the application is essential to being able to reproduce load en masse with fewer than a 1:1 ratio of hosts to number of users tested.

LoadRunner does have a GUI Virtual User, which uses QuickTest Professional (QTP) as the implementation model for this type of virtual user. On a historical basis WinRunner and XRunner, also automated functional testing tools, have been used for this type of user prior to QTP's inclusion. So this is where the tools cross, at the GUI Virtual User which would operate at the top of the application layer of the OSI. For the remainder of the virtual user types (excepting Citrix|RDP and Truclient), you will see that they are spread from the API-middle of the Application layer to the session layer (Winsock).

The higher up the stack (OSI Model) the virtual user maps to then the closer to the GUI and the more expensive from a resource cost per virtual user to execute the test. This is true independent of tool provider. Eventually you can have a virtual user whose cost and resource requirements to execute under load which are no different than a physical user: one virtual user per host. This would then imply a need for 500 computers to perform a 500 virtual user test. Exclusively such tests would be conducted through the front end GUI for an application which was hostile to being run in a shared application server space, such as Citrix or RDP hosting, and whose architecture did not lend itself to grabbing the messages sent from the client to the server. Such environmentally hostile applications are very rare.

Answer 2

in functional testing, we validate a lot. Every UI or API functional test will have a a lot of validation points/checkpoints. In fact one of the key criteria to choose a functional testing tool is to check if it provides rich validation features. If I am not wrong, LoadnRunner or any other performance testing tool is not meant/built to provide these rich validation features. They may provide some basic ones like 'find text' or 'count occurrances'.

Also, I have observed that it takes relatively longer to design a performance test than a functional test.

While understanding it's very tempting to use LoadRunner for functional testing as LoadRunner scripts run faster than UI tests (Selenium/UFT), considering above 2 reasons, I don't recommend it.

Disclaimer: As of March 2020, I only have basic knowledge of performance testing or LoadRunner/JMeter