HLDintermediatePremium

Zoom HLD: Why Video Calls Use an SFU, Not a Mesh

A Zoom / video-conferencing system design: mesh vs SFU vs MCU topologies, an SFU that uploads once and fans out, simulcast layer selection, signaling vs media, NAT traversal, and scaling rooms.

By fiveyearsdevJune 14, 20268 min read

"Design Zoom." A two-person call is easy — connect the two browsers directly. The design appears the moment a third, then a tenth, then a fiftieth person joins: if everyone sends their video to everyone else, each participant's upload grows with the room and a laptop on home Wi-Fi melts at about five people. The whole interview is the topology decision — who sends what to whom — and the answer real conferencing uses is the SFU: every participant uploads their stream once to a server, which selectively forwards it to the rest. Two ideas carry the design: the SFU (one upload per client, server f…

What’s inside

Let's start nowhere near a computer

Where this exact shape shows up

Step 1 — Functional requirements (sentences first)

Step 2 — Non-functional requirements

Step 3 — The topology decision

Step 4 — Simulcast: a layer per viewer

Step 5 — Signaling, media, and getting through NATs

Step 6 — Trade-offs (each one keeping an NFR)

The complete implementation

Step 7 — Only now, the boxes

Step 8 — Scaling the design, one bottleneck at a time

Step 9 — When a piece fails: designing for failure

The interview corner

Where to go from here

Read this one free

HLD