Questions about Jitsi and Jibri servers setup

Hi,

I would like to setup a Jitsi and Jibri server for tele-medicine and also want to record those calls.

My questions are

  1. Let’s say that there are maximum 100 rooms call simultaneously. I need 100 Jibri servers (or maybe 105 - 110 servers to be on the safe side) Am I correct ? If yes, Let’s say I use DigitalOcean to host these servers. So, I need 100 - 110 droplets. Am I correct ?

  2. If I would like to host my own servers at a data center or use other cloud provider that can customize hardware spec. What is a hardware requirement/recommendation for a Jitsi server and Jibri server (each) that have 100 rooms call simultaneously (2 - 5 people in each room) ? Which CPU model should I use ?

  3. According to my use case, If I need a lot of Jibri servers, what is the best way to duplicate a Jibri server ? Installing Jibri to each server is not a good idea, I think. I’ve seen some people use Docker but I can’t find a Docker Image for Jibri anywhere. Any idea ?

Best regards,

You need 100 Jibri instances to record 100 simultaneous meetings. One droplet may have 1 or more Jibri instances according to the hardware capacities and the configuration. If you configure one instance for each droplet then you will need 100 droplet. ~4 cores and ~4 GB RAM for each instance…

I think the best way is to use an auto-scaling system which manages the number of JVBs and Jibris according to the load. It’s very expensive to keep the servers open all the times.

JMS: 8 cores / 8 GB RAM
JVB: 4 cores / 4 GB RAM (many)
Jibri: 4 cores / 4 RAM (many)

I use an already configured image for JVB and Jibri then create the droplet using the same image

How to config a droplet to have multiple Jibri instances ?

There are many way. I prefer to use an LXC container for each Jibri instance