Jitsi easily extends the Docker deployment architecture scheme

Hello, I have recently completed the deployment test through Docker, but it is all stand-alone deployment. In order to deploy a scalable and highly available architecture and put it into production, I have designed a deployment architecture. Please help me to see if it is reasonable to deploy a highly available architecture. Thank you very much!

My indicators:

  1. At most 50 meeting rooms can be opened at the same time, each meeting room has no more than 10 people, and each meeting is automatically recorded through Jibri.
  2. In order to ensure the high availability architecture of the service, a multi machine room and multi node deployment is adopted. The configuration of each VM is RAM: 8G, CPU: 16 cores, Disk: 1TB, and the total number of VMs is 8.

My deployment architecture:

For a single jibri use 8GB of RAM and 4 Cores.

With this architecture, you will basically have half of the jvbs and jibris stay unutilized. If you have a pool of jvbs connecting to both shards and a pool of jibris again connecting to both shards you will be utilizing all of them. Not sure how you can achieve the high availability with docker … maybe have two pools of jvbs and two for jibris and all are utilized all the time by being connected to both shards.

Because only one Jibri instance can be recorded in a conference room, according to my indicator, if 50 conference rooms are opened at the same time, 50 Jibri instances need to be added to the recording. Therefore, my deployment of jvb and jibri is relatively vague. I want to know more about cluster deployment in this regard.