Socket performance testing guidelines
Before you can test the performance of TCP/IP socket-based applications, set up test environment and incorporate these guidelines to produce reliable performance tests.
Limitations
You can use this extension to test applications that run in a client-server model, where the test simulates multiple clients that connect to one or several servers. Other models, such as peer-to-peer networks, are not supported.
The extension does not support applications where the socket DLL is loaded dynamically, such as Microsoft Internet Explorer.
Performance
When deploying performance tests, it is important to use a relevant number of virtual users on a given computer.
For example, if you deploy too many virtual users on a single computer, the results will reflect more the load of the test computer than the load of the server.
For best results with performance tests on an average test computer with a 1 GHz processor and 1 GB of RAM, do not exceed 1000 concurrent virtual users.
If you exceed the number of virtual users that a single test computer can run, the measured performance of the server will be affected by the performance of the test computer, which will invalidate the final results. When editing a schedule for long performance tests, use these recommendations:
- In the schedule editor, reduce the Test log level to None.
- In the schedule editor, set the Statistics sample interval to approximately 1/60 of the run time, for example 12 minutes for an estimated 12-hour session.
- When possible, use loops inside test suites rather than loops in the schedule. Using loops inside test suites avoids connection problems that might occur over long duration tests and emphasizes measurement of the send and receive activity rather than connection and close activity.
Related
Socket API performance testing overview Socket API test editor overview
Related tasks
Record socket tests