0

What is the difference between:

  1. 10r/s and burst=0

  2. 1r/s and burst=9 nodelay

1st case has processing size capacity of 10r/s. So it allows 10 requests to happen in parallel per second, 11th concurrent request is rejected.

In 2nd case the processing size is 1r/s. So if 10 requests are made, then 1 will start executing, and remaining 9 will be added to the burst queue and because there is nodelay keyword, they all will execute immediately. 11th concurrent request is rejected.

Is this correct interpretation?

variable
  • 137
  • 7
  • 1st case allow 1 request every 1/10th of second. If you send 10 request simultaneously only first will be server and other 9 will be declined instantly. – Alexey Ten Feb 16 '22 at 20:50
  • 2nd case allow you to send 10 request at the same time, but 11th request will be allowed only in 1 second. So in a long run first case will allow ~600r/minute, while second case will allow ~60r/minute. – Alexey Ten Feb 16 '22 at 20:52
  • The generic meaning of rate/burst is on this Wiki page: https://en.wikipedia.org/wiki/Token_bucket . Can be applied to a lot of things where there's a rate and a burst to understand them. As for nodelay: https://serverfault.com/questions/247296/nginx-what-is-does-the-nodelay-option-do-when-limiting-requests – A.B Feb 17 '22 at 01:22

0 Answers0