Virtual Users in Performance Testing

Virtual Users in Performance Testing is iterating over a request or generating load on the target system using a programming language by using concurrent threads for requests or by using specialised tool to generate concurrent requests.

In a real scenario, a typical system has consist of user as an actor and software or API interface which provides service to the end user. The end user might directly send request to target system directly (through other API or Software) or through client interface. Generally, a user has one to one mapping with client interface and multiple client interface seeks service from software or API. For example in ticket booking software the user has ticket booking interface on their browser in their device and multiple users are connected to booking system.

In above scenario we need to test the performance of Ticket Booking System, so in this case we need to create load on the target system. Definitely we would not create multiple instances of client interface but by simulating the request which is being sent from the client interface and iterate request through concurrent threads or using specialized tools generate the target load of requests. Many a times people visualise it as performance tools or testing program creating multiple instances of client interface which is not the case.

Determining the number of virtual users depends on the number of projected actual users accessing the system. However, one can observe the system resources with incremental increase in virtual users, starting from small number of users and gradually increase the users by monitoring the system resources and test if system performs as expected with every increase in virtual users. If you needs a formula to calculate number of virtual users one can apply Little's Law.

Conclusion

Performance tests with virtual users are a simulation of requests send by real users and should not be visualised as simulation of multiple interfaces through which real users interact. Optimal virtual users can be determined by monitoring the test with gradual increase of virtual users over time by keeping in mind the actual or projected users of target system.