Analyze the stress test in software testing

One of the concepts of "stress testing" from visualstudio.net Design Distributed Application Reliability testing: is to simulate a huge workload to see how the application performs operations under peak usage. After a stress test for each individual component, stress tests should be performed on the entire application with all its components and support services. Centralized testing starts with the most basic functional test. You need to know the coding path and the user scenario, understand what the user is trying to do, and determine all the ways in which the user applies your application. The test script should run the application according to the intended usage. For example, if your application displays a Web page, and 99% of customers just search the site, only 1% of customers will actually buy it, making it meaningful to provide test scripts that stress-test the search and other browsing features. Of course, the shopping cart should also be tested, but the intended use implies that the search test should account for a large proportion of the test.

The concept of the second "stress test" comes from. NET application Performance test: The stress test is used to assess how the system will run when the maximum load is exceeded. The goal of stress testing is to find bugs in applications under high load conditions. Includes: Synchronizationissues,raceconditions,andmemoryleaks (memory leak). Stress testing allows you to identify the weaknesses of the program and how the program will run under extreme load.

The three "stress test" stress tests are designed to detect changes in the performance of a software system under one (arbitrary) set of conditions. By changing the input of the application to exert an increasing load on the application (concurrency, cyclic operation, multi-user) and measure the performance changes in these different inputs, which is commonly said: The stress test examines the maximum load that the system can withstand in the current hardware and software environment and helps to identify the bottlenecks in the system. This test can also be called a load test, but a load test typically describes a specific type of stress test--increasing the number of users to stress test the application.

There may be more than three definitions of the term stress test described on the web.

I agree with the first concept that stress testing should mean simulating huge workloads to see how the application performs under peak usage. To expand, one stress test should be a short time, followed by a large workload, and again the stress test is to get the application up to a peak. To add to these three points, the 1th-long stress test was transformed into a load test; for 2nd, the pressure on the application was overloaded, so it was constantly pressurized; 3rd, which peaks the application's usage, and if that limit is exceeded, the application crashes or the error rate surges. This peak is for a certain moment, and for a certain critical pressure, the conversion to scenario settings is the maximum number of concurrent users that can be supported.

The purpose of testing in a recent test was to understand the pressure that AUT (the application under test) can withstand, as well as the amount of user access (capacity), and how many users can access a feature at the same time. In AUT, the user's most commonly used five features were selected as the content of this test, including logging in. That's probably what the demand is.

Next i aut login to say how to use LoadRunner and jmeter to achieve the setting of the scene to achieve the purpose of the test. (Note: The server detection is not the focus of this test, this test mainly collects concurrent access to the number of users and the number of errors occurred)