Performing a load test is all about simulation. Its about having an X number of virtual users doing some task on the system and collecting performance counters. This forms a part of non functional test especially around the reliability of the the system.
And so phrases like we need to simulate x requests per seconds will also be thrown about. For instance in the advent of machine to machine communication, a company which has 1000 devices communicating to a load balanced server might need to know just how well the system scales. This can be in line with a vision to have 100,000 devices and hence even consistent with the theory of Enterprise Architecture, it is important to make sure that all aspects of technology supporting the the business are well covered now and in the future.
Additionally the company might want to have a help desk team to respond to user queries. With a total of say 5 help desk agents, then then it needs to know once again just how capable can the system scale with 100 or maybe 200 users. This forms the basis of load testing and brings out the importance of load testing in today’s fast paced technological environment.
The two scenarios described above are similar yet have slightly different approaches when it comes to the Microsoft Visual studio load testing. They are accomplished by setting up a load test whose scenario describes a test mix model of either “Based on user pace” or “Based on sequential test order”. The decision as to when to use which of the two models based on the scenarios I gave are as given below.
Based on user pace – Provides an ability to set the number of tests per user per hour hence consequently being able to set the number of requests a simulator can send per second. This help in achieving the objective of x requests per second.
Based on user pace – With a web test containing a number of steps, these steps are executed sequentially. This is hence the model to be used when simulating user browsing around a website.
Further reading can be found at MSDN.