Learn Large Conference from Zoom Technology

If we read article : https://www.dhirubhai.net/pulse/why-zooms-success-coincidence-distributed-video-yiheng-intel-chen/

So , There are 3 Component make Zoom , can Large Conference - very smooth :)

1. MMR Technology , Beyond MCU
2. Compressing Technology , MBE . Beyond Compression
3. Pagination in Algorithm . Front End handling rendering , which Algorithm for display every participant join , already mapped by Back End in Database / Chace Database

So , if we used Self Hosting Video Conference like Jitsi / Janus

Which technology using SFU , it will not good or drop , if SFU processing reach limit

For Example , in Jitsi , we load test on 1 Video Bridge (JVB) . Core 4 RAM 8 . 1 JVB max handling streaming on 60-70 CCU , or if we count how many streaming , maybe in 60 x 60 = 3600 until 70 x 70 = 4900 (n^2)

It will same , if we used 900 room , each room contains 2 CCU / participant .

1 room = 2 CCU . Stream = 2 x 2 = 4 Stream . If 3600 Stream can handle , so 3600 / 4 Stream = 900 room

So, Jitsi Solution , Like Zoom . If we will create Webinar , with up to max limit from Streaming SFU . It need convert next participant to MCU . Like Streaming or RTMP . Using Youtube / Ant Media , etc

But , we only 1 room 2 CCU with many room , like V-Box (Bank Mandiri)

Video V-Box

It can handle by SFU , Self Hosted - Video Conference Technology , like Jitsi / Janus .

Or in Our Team , We used 3D Avatar for handling Video Conference for Customer - handling with 147 (Infomedia) , Future Depeloyment in All Plasa Telkom and Malls All Indonesia

1 Room - 2 Participant . 147 Agent And Customer

And Already Trial : Implement on Meta Plasa (Virtual)

Like Zoom : 3D and 2D Integration (Zoom + Oculus)

Thank You