What is app performance test
I checked it on the Internet. It seems that there is no special definition. I give my own definition according to my own experience. If there is a coincidence, it is the same.
Client performance test is to design reasonable and effective performance test scenarios from the perspective of business and users, formulate client performance indicators (memory, CPU, number of catons, frame rate, power, loading time, etc.) under each performance scenario, formulate standardized execution processes, execute performance scenarios according to execution standards, and collect performance data with performance testing tools, And analyze the data. If there is a performance problem, locate the problem, cooperate with the development to repair, verify and release, and finally output a complete performance report.
From the above definition, we can conclude that the performance test of app needs to pay attention to the following aspects: the design of performance test scenario, the definition of performance indicators, standardized execution process, performance data collection, performance data analysis, performance problem positioning and performance test report.
Performance testing does not mean that we can find a tool, run a scene, get the data and output a report. Each step should be targeted, so as to reflect the professionalism of testers.
How to do app performance test
Let’s take a look at the following:
Design of performance test scenario
The scenario may be a continuous repetition of an operation or a combination of several operations. For the performance test scenario, it must have repeated operations or continuous operations. The purpose is to enlarge the performance problem to a certain extent through repeated or continuous operations, so that we can find the problem.
Take the tab recommended by station B as an example. If you want to test the performance of feed sliding, the performance scenario can be designed to slide the feed 50 times with an interval of 2S.
Definition of performance indicators
Common performance indicators of the mobile terminal include: memory, CPU, frame rate, number of catons, number of wakp UPS, display duration, etc. what performance indicators we focus on depends on our performance test scenario.
Take the recommended tab of station B as an example. When we cold start the app to enter the recommended tab, we pay more attention to the data display duration, the sliding scenario pays more attention to the number of catons, and we need to seriously consider designing reasonable performance indicators for different scenarios.
Standardized execution process
After the scenarios and indicators are defined, they will be implemented. Standardized implementation is required here. Standardized implementation is not simply implemented according to the definition of the scenario, but needs a lot of attention.
Which specifications can be defined:
-How many seconds does the scenario need to wait before execution
-How many seconds to wait after execution
-Do you need cold start or reinstallation for each test
-How long will it take for the installation to start testing
-Does the test account, test data, equipment and network need to be fixedEvery point may affect the accuracy of performance data. Specifications must be defined and implemented every time. Moreover, this specification is dynamic. With our continuous testing, we will find many problems affecting performance data, which must be customized and avoided. At the same time, good specifications can lay the foundation for performance data analysis later.
Performance data collection
Performance data collection may be the simplest part of the whole client performance test. There is a mature tool perfdog that can be used. It is convenient and simple. You can also use the commercial perfdog service to realize automatic performance data collection, which requires money.
Performance data analysis
After collecting the performance data, we need to analyze the data and how to analyze it. Let me briefly talk about it. An article will be published later specifically on how to analyze the performance data
-
Trend chart. From the trend chart, we can roughly see the performance of the scene in the current version, and we can draw the following conclusions:
- Compared with the previous version of the trend chart, the fluctuation of performance indicators
- Performance index peak value, scenario mean value and increase change
- The starting value of the scene is different from that of the previous version
- The value after the end of the scene changes from the previous version
Performance problem location
After performance data analysis, if there is a problem, you need to locate the problem, which business problem or which Mr caused the problem, and you need to go back.
- First find the developer and communicate with the developer to see if the problem can be determined according to the problem appearance. If it cannot be confirmed, it is necessary to test which Mr combination caused the positioning
- List all Mr’s included in this version, and filter out those Mr’s that are the business of performance problems
- Find the package before and after Mr closing and run again to confirm whether each MR has an impact
- After determining which Mr integration caused the performance problem, communicate with the developer again
Performance test report
The purpose of the performance test report is to give the performance of the current version, which needs to include some core modules
- Test conclusion
- Performance problem attribution
- Performance index data of each scenario
- Test environment and scheme
- Performance index trend chart of each scenario
I have some superficial understanding and experience of APP performance testing. If you have any questions, you can leave a message and discuss them together..