Samvyo blogs

Concurrent Video Session Scale: What Has to Hold at Org-Wide Peak

Steady-state capacity is not the test. The test is 8,000 people joining in four minutes, either into one all-hands or into 2,000 small meetings. Same headcount, different bottlenecks, and the first thing to break is usually the join path, not the media. Here is the arithmetic.

Video Conferencing Infrastructure

SFU Scaling: Clustering, Cascading, and Where the Bottleneck Moves

Scaling an SFU is not one problem. The bottleneck moves three times — from the worker process, to the network interface, to the links between servers — and each move changes what you should be measuring. Here is where each limit bites, with the arithmetic behind it.

Video Conferencing Infrastructure

Scaling Video Cost: What Happens When One Media Server Isn't Enough

A single media server does not fail at a user count. It fails at an egress number, and that number grows with the square of the room. Here is the arithmetic behind a 40-person call, the bandwidth bill it generates, and why the same workload costs 90x more on one venue than another.

Subscribe to Samvyo blogs