Wowza low-latency: LL-HLS, WebRTC delivery, and the latency profiles
Wowza's low-latency live capabilities — LL-HLS, WebRTC delivery, hybrid latency configurations, and the realistic end-to-end latency profiles for each.
Wowza supports multiple low-latency delivery modes — Low-Latency HLS (LL-HLS), WebRTC delivery, and hybrid configurations. For operators evaluating live latency, the right mode depends on the latency target: 4-8 seconds (LL-HLS), 1-3 seconds (WebRTC delivery), sub-1-second (specialized WebRTC configurations).
What Wowza actually has
LL-HLS support with CMAF chunked transfer encoding. Sub-second segment chunks (typically 500-1000ms) emitted via HTTP/2 server push. WebRTC delivery via Wowza's WebRTC streaming targets — playback in browsers via standard WebRTC peer connection. Hybrid: ingest as RTMP or SRT, deliver as both LL-HLS (for cacheable mass distribution) and WebRTC (for low-latency interactive subset of viewers). CMAF chunked output works at the packager layer; downstream CDN must support HTTP/2 push or HTTP/3 stream priorities for the latency math to land. WebRTC delivery scales differently than HLS — peer-to-edge connections per viewer rather than CDN-cached HLS.
Where it's the right fit
Workloads with mixed latency requirements (a small interactive subset of viewers wants WebRTC sub-second; the broader audience is fine with 4-8 seconds via LL-HLS). Live commerce, auctions, betting platforms where sub-second latency is required for the moment of decision. Educational platforms with interactive Q&A elements during live broadcasts.
Where the gaps show up
WebRTC delivery at very large scale (50K+ concurrent viewers per stream) is more expensive than HLS — the per-viewer connection model doesn't cache. LL-HLS sub-3-second latency requires HTTP/2 push support throughout the delivery chain; some CDNs lag here. Latency optimization across the full chain (encoder → packager → origin → CDN → player) requires expert-level tuning that's not push-button.
Pricing implications
Low-latency support is included in Wowza Streaming Engine and Streaming Cloud at no additional charge. WebRTC delivery scales with concurrent viewer count more aggressively than HLS — at large audiences, the operational cost can dominate.
MpegFlow Live (2026 Q3) ships with LL-HLS via CMAF chunked output as the primary low-latency mode — covers the 4-8 second latency band. WebRTC delivery is on the 2027 roadmap; for sub-second latency today, Wowza or specialized WebRTC vendors (Daily.co, Agora, Twilio) are the production choice. The honest assessment: WebRTC delivery at scale is its own discipline; we're focusing on the LL-HLS sweet spot first.
- low-latency
- wowza
- ll-hls
- webrtc