Several key indicators in web performance testing: number of concurrent users, QPS, average request waiting time for users

Regarding the number of concurrent users and QPS, I have been entangled by these two concepts. After reading the relevant information, the summary is as follows: The number of concurrent users and QPS are not directly related, but if you want to talk about QPS, you must specify how many concurrent users are. QPS under the number of users, otherwise it is meaningless, because 40QPS under the number of single users and 40QPS under the number of 20 concurrent users are two different concepts. The former shows that the application can serially execute 40 requests in one second, while the latter shows that in the case of 20 concurrent requests, the application can process 40 requests in one second. When the QPS is the same, the larger the number of concurrent users , which represents the better the concurrent processing capability of the website. For the current web server, it must be more than enough to process a single user's request. At this time, there will be a waste of resources (on the one hand, the server may have multiple CPUs, but only process a single process, on the other hand, when processing a process , some stages may be IO stages, which will cause the CPU to wait, but whether there are other requesting processes that can be processed). When the number of concurrency is set too large, there will be many requests to be processed every second, which will cause frequent switching of processes (threads). At the same time, the user's request waiting time will also increase, even exceeding the user's psychological bottom line. Therefore, between the minimum and maximum concurrent numbers, there must be an optimal concurrency value. Under the concurrent number, QPS can reach the maximum. However, this concurrency is not an optimal concurrency, because the concurrency when the QPS reaches the maximum may have caused the user's waiting time to exceed its optimal value, so for a system, the optimal number of concurrency must be It needs to be comprehensively determined in combination with QPS and the user's waiting time.