Article From:https://www.cnblogs.com/byron0918/p/9061047.html

Transfer from: https://www.cnblogs.com/xianlai-huang/p/7795215.html

JmeterThe calculation of the Throughput and the average RT

1.TPS:The number of transactions per second, the Throughput of JMeter is the throughput (request number / sec), and after the transaction controller is added, TPS=Throughput

Macroscopically: TPS= concurrent number / response time, JMeter Throughput = (number of requests) / (total time), i.e.

Throughput =(sampleSample number) / (the last thread starts time + the last thread’s last time – the first thread starts time).

It is possible to understand this formula: absolute concurrency does not exist, the time of the request is always successively, and the absolute TPS can not be calculated. In a statistical perspective, the total number of requests for the server processing requests is TPS, which is also the same.

Why do we need to increase the number of users to find the largest TPS of servers?

2.Average response time = the response time and sample number of all sample samples.



TPS=1/RT*The number of users (RT is the average time of all things)

It’s wrong.

If 1, the A application is a single thread processing, processing a request requires 1s, 5 VU to request, the first request cost 1s, second spent 2S… Fifth spent 5S, the total time is 5S, the total request number is 5, so TPS=1, the average response time is (5+4+3+)2+1) /5=3s, which is incorrect according to the TPS=1/RT* user number at this time.

If 2, when we ask for many things, this formula is obviously wrong.

Leave a Reply

Your email address will not be published. Required fields are marked *