Chapter 5 LoadRunner performance testing practices

Chapter 1 Performance Test Result Analysis

Performance analysis has always been a problem, but I think it is difficult to design and execute.
This article is the introduction to Chapter 5 of LoadRunner performance testing.
In chapter 5, I want to avoid tools and discuss performance testing and analysis in order to better understand the problem. Therefore, let us share some content for your discussion. This chapter is being written .......

After the test scenario is completed, many test engineers think that the most difficult stage is coming-performance test result analysis. Therefore, this chapter naturally becomes the most important chapter. However, the author has always believed that performance test analysis is not the most difficult task. The performance analysis is also dependent on the design and implementation of these foundations. It can be said that the performance test analysis is only the last part of the competition. 20 meters . Of course, this does not mean that performance test analysis is not important, because "the last 20 meters "The result of the previous work is futile. It is hard to understand that the foundation of performance testing and analysis is the test result data executed in the previous test scenario. To ensure that the performance test and analysis conclusions are correct, the test result data should be correct, which means that the test scenario and test execution process should be correct.

In fact, performance testing should be a very rigorous project from the beginning, and the work in each stage is closely linked. Performance Testing engineers should take the work of each stage seriously. If you are blindly seeking to find out the system bottleneck, it is undoubtedly a case-by-case approach.

If this chapter disconnects from practical application or provides some isolated test results to explain how to use LoadRunner's analysis, many readers will be confused and still cannot solve the actual problem. Therefore, this chapter first explains how to analyze the performance test results based on cases, and then discusses the specific use details of analysis. This chapter is different from chapter 3 and chapter 4.

The main content of this chapter is as follows:

L analysis performance analysis basics

L how to analyze some typical analysis charts

L use cases to explain the performance test and analysis process

5.1 how to analyze performance test results

After the test scenario is completed in the controller, you must first determine whether the collected result data is true and valid. Most performance test scenarios require repeated and iterative tests. Therefore, many test results cannot reflect the problem themselves. In-depth analysis of such results is a waste of time. This book mainly discusses how to analyze the effective test result data.

To determine whether the test result is valid, follow these steps:

Step 1: Check whether the test environment is normal during the entire test scenario.If some exceptions occur during the testing process, such results are often inaccurate and do not need to be analyzed.

For example, during the test execution, the CPU usage of the test machine often reaches 100%, the network in the test environment is unstable, and some system parameter configurations are incorrect. Such test results do not need to be analyzed, you should reset the test scenario or adjust the test environment to execute the test again.

Step 2: Test whether the scenario settings are correct and reasonable.The correct setting of the test scenario has a great impact on the test results. Therefore, after a test, you need to analyze whether the scene settings are incorrect.

Some new users may load all virtual users on one PC at the same time when using the Controller for testing. For example, if the client cannot process 1000 virtual users at the same time, some virtual users fail because they cannot be initialized. These failures are not handled by the application server to be tested, but are not transmitted by pressure. In this example, the correct method is to add more load generators or gradually pressurize them to continuously try to run the test scenario.

Step 3: Check whether system problems are exposed.During the execution of the entire test scenario, there is no need to analyze all the normal results of the system, because such results do not reflect system performance problems, and further adjust the scenario (such as increasing the pressure) test. The results generated in testing scenarios that make the system abnormal during the testing process need to be thoroughly analyzed. In fact, the testing results that reflect performance problems are the main tasks in the performance analysis stage,

The test results expose many system problems, for example, during the test, the user transaction response time is too large, the maximum number of concurrent users supported by the system is too low, the CPU utilization of the system application server is too high, or the memory is insufficient. For such test results, the performance tester needs to use analysis to conduct in-depth analysis on the performance test results to find some potential performance problems.

 

This section first introduces the basic knowledge of performance testing and analysis, and then introduces the basic knowledge of LoadRunner analysis. It is best to introduce how to identify and solve system performance problems based on cases.