Information by project: Server requirements for 1000 concurrent calls 1 - 1 ( 1 Partecipats)

Hi everyone. I apologize for the bad english.

We are thinking of realizing a project located mainly in Italy, in which we would like to implement jitsy to provide customer assistance within a virtual tour.

And here I come to the questions:

We would like to know what kind of infrastructure at the level of hardaware and bandwidth network we should have to support 1000 simultaneous calls with 1 participant per call only in addition to the moderator, to give helpdesk assistance with meet to individual customers.

I read that now with the new videobridge2 it is possible to scale the infrastructure

I have read about the strict limit of 75 participants per call, but it is not our case we are interested in knowing how to support 1000 simultaneous calls from 1 participant over the moderator.

Thanks in advance

1 Like

Hi. I have a setup for 8k simultaneous users and working fine (meetings of around 15-25 participants each). I recommend to setup a server with jitsi-meet using the quick start (https://jitsi.github.io/handbook/docs/devops-guide/devops-guide-quickstart) then install jitsi videobridge in other server (https://github.com/jitsi/jitsi-meet/wiki/jitsi-meet-load-balancing-installation-Ubuntu-18.04-with-MUC-and-JID). Then, you have to make sure that all is working fine, and my recommendation is to stop the videobridge in the jitsi-meet server and use it only as a load balancer.
After that, when you are sure that all is working, you can setup an autoscaling setup, creating an image of your jvb and making an startup script to change the muc_nickname of the newly created servers of Jitsi videobridge. What I use is:

#!/bin/bash
sed -i -e “s/YOUR_IMAGE_MUC_NICKNAME/$(uuidgen)/g” /etc/jitsi/videobridge/sip-communicator.properties
systemctl restart jitsi-videobridge2.service

Specs of the servers, is enough a jitsi-meet server of 4 cores and 8 or 10GB of Ram and a bandwith of at least 1gb/s and use dedicated vcpus or compute optimized servers.
With an autoscaling setup you will save money and get a good performance, it’s better to have a lot of small servers than a hughe server that is going to fail because of bandwith issues. Remember just put what you think is the minimium of videobridges to suport your normal usage and then put an autoscaling “aggresive” policy if there is the need to launch more instances, and a conservative policy to terminate those instances.

1 Like

Sorry for the delay, thanks a lot for the moment… now I try to implement the infrastructure and let you know