WebRTC bundles several IETF specs into one cohesive browser API. The transport is UDP with DTLS-SRTP for encryption; connectivity uses ICE with STUN and TURN; media is RTP and feedback is RTCP; signalling (SDP offer/answer) is done by the application over its own channel (typically WebSocket). The API surface — RTCPeerConnection, getUserMedia, RTCDataChannel — has been a W3C Recommendation since January 2021 (W3C "WebRTC 1.0").
In streaming, WebRTC has two distinct roles. Peer-to-peer (the original use case) connects two browsers for video calls — Google Meet, Zoom Web, Whereby. Egress (server-as-source) uses WebRTC to deliver one stream to many viewers via an SFU — Twitch's low-latency mode, Kick, sports betting platforms. Both roles share the same protocol stack but have different scaling characteristics.
WebRTC's strength is sub-second latency and ubiquitous browser support. Its weakness is operational complexity — STUN/TURN deployment, SFU scaling, congestion control tuning. The 2022–2026 trend is to wrap WebRTC behind simpler protocols: WHIP for ingest, WHEP for delivery, Media over QUIC as a longer-term successor. Cloud providers (Cloudflare Calls, Mux, Dolby.io, LiveKit Cloud, AWS) sell managed WebRTC to hide the complexity. Self-hosted WebRTC remains viable but demands real expertise.
