How to test performance

Considerations for preparing performance tests in advance:

1. The environment configuration for performance testing should be able to simulate on-site use of versions as much as possible, including internet devices, software network elements, various hardware platforms, operating systems, and software platforms;

2. For performance testing, appropriate simulated scripts should be prepared to simulate possible operations of the customer as much as possible, such as parallel webpage operations and socket connections at the same time. In addition, it exceeds the real conditions of the customer.

Two types of data are required for performance testing:

1. Benchmark Test comparison data: Compare the performance indicators of the current and previous versions. It is used to identify whether the integration of functions of the current version affects the benchmark performance. In the case of benchmark testing, the new features and features of this version are not enabled by default, and they must be consistent with those of the previous version.

2. Performance comparison data of a single function: verify the performance impact of this function on the version when new features and features are enabled in the current version.

Performance testing focus:

1. resource usage: View resource usage. Resources include CPU, memory, and hard disk.

2. Resource release: whether the system can normally release resources after business processing is stopped for future business use. When the business stops, resources should be released in a timely manner. Common problems: Memory leakage and resource suspension make the system unable to release resources normally. Serious problems cause downtime. You can use many tools to check resource conditions.

3. Exception test: if the performance test is performed in certain cases (typically by simulating on-site users), hardware switching, dual-host switching, and service switching are performed. This includes destructive input access to verify the system's fault tolerance under high loads.

4. query alarm and other information: Generally, the system will send notifications and alarms when a problem occurs. This information is the best way to expose the problem. You must check the performance test in time.

5. long running: performance testing is a good way to simulate long running of devices. It can detect many problems related to accumulation effects such as time and timer.

6. Log check: performance tests require frequent analysis of system logs, including operating systems, databases, and software versions.

7. view the business response time: after a long test, check whether the business response is within the acceptable range of the customer. For example, the webpage response time and terminal Logon Time.

Performance testing personnel requirements:

1. Performance testing personnel must be the backbone and cannot use new users for performance testing.

2. Performance testing personnel must be familiar with the entire system and be proficient in problem locating. Lead developers to troubleshoot performance problems.

Performance test report:

1. The performance test report should reflect the benchmark performance data and performance data of a single function. It is used to assess whether the version can maintain the same processing capability in the original hardware environment.

2. The performance test report must meet the requirements of various test stakeholders. Therefore, before performing a performance test, you need to obtain the requirements of the testing stakeholders, make a detailed list, and then start the performance test.

Performance testing tool requirements:

1. Prepare certain tools for performance testing, including LR. Performance testing of many products requires self-developed performance testing tools. The highest level of the tool is to fully simulate customer operations. In particular, LR is just a tool, and performance testing is a set of theories and methods.

2. manual operations are required during the use of the performance testing tool. For example, simulate online shopping activities of customers. Tools and manual tools must be effectively combined. To make up for some unpredictable shortcomings of the tool.

Performance testing is the key point of system-wide testing. We need to pay great attention to testing design, testing execution, and personnel arrangement.

How to test performance