How many simultaneous participants for 1Gb/s on JVB?

Hello everyone,

I have 1Gb/s of available bandwidth IN and 1Gb/s OUT on JVB server and I would like to know how many participants could JVB handle .

Here are my calculations without considering simulcast (Participants with Firefox only) :

All users sending 720p resolution :
IN on JVB : 4 Mbps/user
OUT on JVB : 4 Mbps/user
Number of simultaneous participants : 1024/4 = 256

All users sending 480p resolution :
IN on JVB : 2.66 Mbps/user
OUT on JVB : 2.66 Mbps/user
Number of simultaneous participants : 1024/2.66 = 384

Are these numbers correct ?

Thanks in advance.

This is not correct as it depends on the sizes of the conferences and the output varies.
So let’s say you have 6 participants - the jvb IN will always be 6*4=24.

  • 1 conference with 6 people, every one receives 54 which is 65*4=120
  • 2 conference with 3 people, every one receives 24 which is 32*4=24 for two conferences total make 48.
    6 in one conf out is 120, 6 in 2 conferences out is 48.
1 Like

It depends on many things: conference size, whether p2p is used and whether you have a TURN server configured, what browsers are used (we don’t have simulcast on firefox), how often video is muted, and probably other variables as well. On meet.jit.si we see an average of about 1-2Mbps per user combined (upload + download).

Regards,

Boris

1 Like

Thank you @damencho, your calculations are correct.

Indeed, it’s not easy to get an exact number.
We don’t use P2P nor TURN. The majority of our clients use Firefox

Thanks guys!

We have tested with near to constant upload speed and it is minimum 2.5Mbps per user constant speed is required. For basic and lowest resolution to work flawlessly, it is 3Mbps required in general.
Download data used is almost 1/2 of upload as of TOTAL CONFERENCE with 3-4 users.

So I would recommend min. 3Mbps would be delivered per user for upload + download.

However I have not tested with Audio Only.

What is required for a single conference and what you see in practice is different. What I quoted is the total traffic divided by the number of users.

Boris

yeah I agree