Comparing MOQ with Low-Latency HLS and DASH
How does MOQ stack up against Low-Latency HLS (LL-HLS) and Low-Latency DASH (LL-DASH)? A technical comparison of architectures, latency profiles, and trade-offs.
The Low-Latency Landscape
As live streaming demand has grown, both Apple (LL-HLS) and DASH-IF (LL-DASH) have introduced low-latency extensions to their respective protocols. MOQ represents a more fundamental rethinking of the media delivery stack. Let's compare them.
Architecture Comparison
LL-HLS / LL-DASH
- Built on HTTP/1.1 or HTTP/2
- Segment-based: media is divided into segments (typically 2-6 seconds)
- Uses chunked transfer encoding and partial segments for lower latency
- Relies on standard CDN infrastructure
- Manifest/playlist driven (M3U8 for HLS, MPD for DASH)
MOQ
- Built on QUIC/WebTransport
- Object-based: media is divided into fine-grained objects
- Uses QUIC streams for multiplexing and prioritization
- Requires MOQ-aware relays (new infrastructure)
- Catalog-driven track discovery
Latency Comparison
| Protocol | Typical Latency | Theoretical Minimum | Glass-to-Glass |
|---|---|---|---|
| HLS | 10-30s | ~6s | ~8s |
| LL-HLS | 2-5s | ~1.5s | ~2s |
| LL-DASH | 2-4s | ~1s | ~1.5s |
| MOQ | 100-500ms | ~50ms | <500ms |
MOQ achieves significantly lower latency because:
- No segment boundary waiting — objects are delivered immediately
- QUIC's 0-RTT connection establishment
- No manifest polling overhead
- Stream multiplexing eliminates head-of-line blocking between tracks
Scalability
All three approaches can scale to millions of viewers, but through different mechanisms:
- LL-HLS/LL-DASH: Leverage existing HTTP CDN infrastructure. Every CDN in the world can cache and serve segments.
- MOQ: Uses a relay cascade model. Requires MOQ-aware relays, but these can be integrated into existing CDN edge nodes.
Adaptive Bitrate
- LL-HLS/LL-DASH: Client-driven ABR. The player chooses quality based on network conditions and switches between renditions.
- MOQ: Track-based switching. Subscribers can switch between quality tracks at group boundaries. The relay infrastructure can also assist with quality selection.
Migration Path
For existing streaming platforms, the migration path matters:
- Short term: LL-HLS/LL-DASH for improved latency with existing infrastructure
- Medium term: MOQ for new real-time features alongside existing streaming
- Long term: MOQ as the primary delivery protocol for both live and interactive use cases
Conclusion
LL-HLS and LL-DASH are pragmatic improvements that work today with existing infrastructure. MOQ is a generational leap that requires new infrastructure but offers fundamentally better latency and real-time capabilities. The two approaches will likely coexist for years, with MOQ gradually taking over latency-sensitive use cases.