When you perform stress testing, you set the number of requests per second (RPS) to 150 to limit the maximum number of transactions per second (TPS). However, the stress testing data indicates that the TPS value is 192, which is greater than the configured RPS value.
The preceding issue occurs because the response time (RT) of each request varies and the TPS data collected by the client is the total number of transaction responses received by the client in the previous second. In the preceding example, at least 42 requests are not responded to within 1 second. This number is calculated by using the following formula: Current TPS value - Configured RPS value = 192 - 150 = 42.
In the following figure, the RPS value is set to 5.
If the RT of the five requests sent at the beginning of the stress test is 1.5 seconds and the RT of the five requests sent in the 1st second of the stress test is 2.3 seconds, the TPS value in the 1st second is 5. This indicates that the five requests sent at the beginning of the stress test received responses within the 1st second.
If the RT of the three requests sent in the 2nd second is 1.5 seconds and the RT of the other two requests sent in the 2nd second is 0.5 seconds, the TPS value in the 2nd second is 2. This indicates that the two requests sent in the 2nd second whose RT is 0.5 seconds received responses within the 2nd second.
If the RTs of the five requests sent in the 3rd second are 0.5, 0.8, 1.2, 1.5, and 2.1 seconds, the TPS value in the 3rd second is 10. The TPS value includes the five requests sent in the 1st second and received responses after 2.3 seconds, the three requests sent in the 2nd second and received responses after 1.5 seconds, and the two requests sent in the 3rd second and received responses after 0.5 seconds and 0.8 seconds.
