How to Write Performance Test Cases

Because performance tests differ greatly from functional tests, the results discussed may differ from the previous assumptions.
Purpose:
In order to verify whether the system meets the performance indicators proposed by the user and detect the performance bottlenecks in the system, the system is optimized.

Source of performance testing indicators:
The user's explicit requirements on various indicators; if the user does not propose performance indicators, design various testing indicators based on the user's needs and experience of the test designer. (Requirement + experience)

Main Performance indicators:
Server metrics (CPU and memory usage), backend database metrics, network traffic, and response time.

Bug View:
1. Performance Testing is like running without any wind (performance indicators under normal circumstances );
2. stress testing is like running in the breeze (performance indicators that increase the percentage of stress on a normal basis );
3. Load Testing is like running in a strong wind (continuously pressurization until the system crashes ).

HTTP opinion:
1. Load Testing is a continuous pressurization under normal circumstances;
2. The pressure test is to directly Pressurize to reach a limit value.

A unified view:
Performance testing, stress testing, and load testing are inseparable and can be collectively referred to as performance testing.

Key Points of performance testing:
1. The performance test is performed after the function test is completed.
2. The performance test plan and solution are generally in the same document as the test case.
3. The test environment should be consistent with the user environment as much as possible.
4. Generally, performance testing is completed by testing tools and testing personnel. The performance testing environment should be run independently to avoid simultaneous use with other software.
5. Performance Testing focuses on the design of preliminary data and analysis of later data.
6. The performance test cases mainly involve the entire system architecture. Therefore, once the test cases are generated, the changes are generally not significant. Therefore, the repeated usage of performance tests is generally high. (Note: When a function point in the system needs to be modified, it generally only affects the design case of the function test. For performance test, it rarely affects the design case of the performance test. However, if a function has been greatly modified, the performance test should also be retested .)