Performance test web articles
Author: piaocl
Keywords: performance test, web test
Summary:
With the rapid development of the online world, the performance of the website has become increasingly important, and the performance of a bad performance will be abandoned by the user. So performance is an important aspect of the user satisfied with the software system. This article will discuss what is performance, how to test performance, etc.
So what is performance, performance is the time and resources of a particular function. He can be a function overhead or the number of synchronous run functions. The web performance test is to simulate a large number of users to cause pressure to the website, and evaluate the WEB system to meet the standards that have been defined under different loads and different configurations. Performance tests are more concerned about analysis and eliminate performance bottlenecks associated with software structures.
Performance is an indicator that must be considered for each software system. We usually pay attention to the following four data in performance test: III. Data traffic three .. The software itself consumes four. System usage. Due to the particularity of performance testing, in general, a special test tool (such as LoadRunner, TestManager, ACT, etc.) simulates multi-user operations, which requires pressure on the system that needs to be evaluated. Find the bottleneck of the system and submit it to the developer for amendment. So the purpose of performance test is to identify system performance bottlenecks and correct issues that need to be corrected.
Developing a Web Performance Test that can follow the most important modules in the system, often used modules, the most important part of the system overhead (the most complex part of the code), test the user's most important part of the user. We analyze the performance of performance may have problems from the overall perspective. As a web program, no matter two layers, or a multi-layer system architecture program. The system is through the HTTP protocol (Hypertemx Transfer Protocol, HTTP) is a "Request-Response" application protocol, which supports a fixed method such as GET, POST, PUT, DELETE, etc.) to transfer data. ,send command. The middle process is approximately to the client (browser) to send a data operation request to the server, and the server receives the command request to process, and then returns the result to the client, the client process the response result. In this process, the process of data processing in this process is generally not a system performance bottleneck. This time is generally a large number of users simultaneously, and the sending data request is the bottleneck of the system. The system's server hardware resources are most likely to be bottlenecks of software systems. We need to adjust software and environment (optimized response time and resources), confirm the high load and pressure conditions that apply and system can handle. And this time the system hardware is the bottleneck caused by the CPU. The bottleneck caused by other hardware resources in the system, we can replace the corresponding hardware resources and adjust it. Another bottleneck in the system may be that the server sends processing data to the client. This process hardware resources are impossible to be performance limits, most likely the reason for bandwidth.
In fact, the problem of performance is due to software architectural or design unreasonable, rather than coding. If the CPU usage is high in performance test, the memory occupies an abnormality. When the situation of the abnormality often reports, the reason for the on-side analysis should be submitted to the developer, so that the developer checks the code in the code. Logical errors such as dead cycles, check if the data connection driver used after application deployment is incorrect or the code is not updated in time, check if there is an abnormality, interrupt test, only run the longest response time, and verify results . Analyze system possible performance bottlenecks.
We need to pay attention to the results of different bandwidth conditions on test results, so we should formulate system performance standards before performance testing, so that test results should be reasonably analyzed. The test results of different bandwidth were analyzed to find the test system bottleneck.
The most common basic type in general performance test is a reference test, configure test, load test, pressure test, and we know that during software testing, different stages, different types of performance test targets are different, different software The architecture also determines that there is a difference in performance testing. This will have a certain understanding of the type of tests that can be better enabled by better performance testing. The following is the specific concept of various test types: a reference test - compares the performance of the new server or the performance of the server and known reference criteria.
Configuring Test - Confirm that the server is acceptable under different configurations. (The operating conditions are unchanged)
Load Test - Confirm the acceptability of the server under different load conditions. (The operating conditions are unchanged)
Pressure Test - Confirm that the server is acceptable when the server is in an abnormality or limit condition, for example, reducing resources or a large number of users.
Competitive Test - Confirm that the server can handle a number of clients to compete for the same resource.
The baseline of how the server processes the data under certain conditions provided by the system reference test, as the starting point of the reference data for evaluating other performance indicators. Configuring tests is whether the test system is configured to function properly on a different machine. Use configuration tests to ensure that the system runs normally on multiple platforms. The load test is used to test the response time of the client or server side under different load conditions. Helping testers calculate the maximum number of transactions that are requested in the qualified time server response processing. The pressure test is the process of running the system under extreme conditions, checking what the server or client crashes under the condition. Competitive tests are operating system functions on one or more PCs to simulate the actual environment.
Some testers may face a released web system, which is how to perform performance testing. You can collect test data through the management personnel of the website, and prepare for performance testing, the information collected includes the number of online users per hour, the average number of online users, online user peak comparison Collect the number of pages that have been accessed, and the usage rate of the CPU is running normally. All of these as the basis for performance testing to avoid blind performance tests. It is likely to be a wrong test without understanding the real situation of the system. Put too much energy in the wrong place. The test results do not have authenticity and waste a lot of test resources.
For performance testing in a development of the Web system, it is necessary to constantly adjust the test target based on the development goals and user needs, constantly adjust the test target, and have a good assessment of system performance. So how can I successfully perform performance testing, I think there should be a test plan before performing performance testing, so that performance test is undressed in order. The following graph gives the general steps of performance test:
Figure 1-1 Performance Test Procedure
n Evaluation System: At this stage, it is mainly to clearly determine the system expectation target, including: determining system functions, determining user activity, determining system architecture, determining acceptable limits, verifying acceptable limits, determining system risk, etc.
Develop test resources: Resources for performance testing, mainly covering three activities: development risk reduction plans, development test strategies, and develop automation scripts.
n Performing a Benchmark: Use the baseline to compare the future test results to the system to evaluate load, pressure, and competitive test results. The correctness of the automated test script is mainly verified by the pressure scene. The visible performance indicator is marked in the early test cycle. Performance projects include individual script baselines to establish initialization benchmarks.
n Analysis Test Results: Analysis of the results is to determine whether the system will continue to test, whether the result reaches the expected value. This stage contains the results of the assessment, determines if it is acceptable, determines whether the test is continued, determine the demand adjustment.
n Book Test: Force verification system performance, these tests must be performed even if performance and demand are not adjusted. It contains activities that perform user experience tests, perform stability tests, and perform product confirmation tests. n End Test: Due to software architecture and other resource restrictions, software cannot continue to optimize due to software architecture and other resource restrictions. At this time, it is recommended to transfer test resource to other respects to avoid waste of test resources.
Through a series of series, everyone should have a certain understanding of the purpose of Web performance testing. There are thousands of differences in actual work, and you need a specific problem to analyze to do performance testing.
references:
1. Performance_ENGINEERING_STRATEGY_TEMPLATE
2.
