Replies: 1 comment 2 replies
-
Hi @Tsukiand Capacity is a very elusive topic... It somewhat depends on the use case. If I understand correctly you want to know what is the upper limit for a specific machine. I suggest following for you to try: overload a single instance of coturn with traffic as close to production as possible and see where it stops scaling with traffic. This could be done with test traffic (for example generated by turnutils_uclient) or some other test infra if you have one or actual production. What closely following metrics (not coturn exposed necessarily)
Once you see one of the flatten out (or your CPU getting to 100%) that is when you know you hit some limit. This is your "100%". From that knowledge you can figure out what is the limit and how you can use it. |
Beta Was this translation helpful? Give feedback.
-
We have coturn cluster but we have no idea how much capacity it used. We just use network throughpu and udp packet loss rate as an indicator. We want to have a clear data to describe the capacity of coturn server.
Beta Was this translation helpful? Give feedback.
All reactions