Performance Testing
The purpose of performance testing is to determine changes in the application, depending on the intensity of the load.
For example, in the first 20 minutes simulate the simultaneous operation of 10 users (10 users perform the operation at intervals of 5 seconds). Then the next 20 minutes are increasing number of users to 20, following 20 minutes – increasing to 50 i.t.d until the application does not fall down or until you start working so slowly that the work can not call it.
Note: The number of users and the range depends on the system under test. Some systems for normal operation is sufficient to support simultaneous operation of 10 users, and some – 10 thousand users.
You can also change the interval operations.
The more users – the greater the load on the system.
At intervals all the way around, the smaller the range of operations – the more load on the system.
For performance testing we are interested in:
- change in the run-time operations, depending on the intensity of operations (where operations intenvisnost = number of users * number of operations per unit time)
- definition of the limits of acceptable performance (where an acceptable performance – either explicitly prescribed in the TOR the average response time, or if the speed when the application is normally impossible to work) By the way, for different operations, acceptable performance may be different. For example, if a user application to perform the operation 10 times per hour – then it must be performed quickly and does not irritate the user’s eternal waiting. If the user needs to perform the operation once a week or even once a day (for example, to generate complex reports), then such an operation can be carried out even longer. User at this time may do other things or just go to drink coffee.
- determine the number of users that can simultaneously work with the annex.
See Also