What are Streaming Protocols and the Ingest vs Delivery Architecture?

Definition
Streaming protocols define how video and audio data is transmitted from source to viewers. They split into two categories: ingest protocols (broadcaster to server) optimized for reliability, and delivery protocols (server to viewers) optimized for scale.
Ingest Architecture
Ingest moves live video from the broadcaster (camera, encoder, or streaming software) to the streaming platform. RTMP (Real Time Messaging Protocol) dominates ingest because it provides 1-3 second latency, runs over TCP for reliable delivery, and has universal encoder support. The broadcaster sends a continuous stream to an edge PoP (Point of Presence, a CDN server location). Once ingested, the stream is transcoded (converted into multiple quality levels) and repackaged for mass delivery. Alternative ingest protocols include SRT (Secure Reliable Transport) and RIST (Reliable Internet Stream Transport), which use UDP with ARQ (Automatic Repeat Request, a mechanism that retransmits lost packets) for better loss recovery on unreliable networks.
Delivery Architecture
Delivery distributes video to millions of viewers using HTTP-based protocols. HLS (HTTP Live Streaming) and DASH (Dynamic Adaptive Streaming over HTTP) dominate because they work over standard HTTP ports (firewall-friendly), leverage existing CDN infrastructure, and support ABR (Adaptive Bitrate) streaming where players automatically switch quality based on network conditions. The separation of ingest and delivery enables independent optimization: ingest for reliability and speed, delivery for scale and cacheability. A single ingest point fans out to millions of viewers through CDN distribution.
Scale and Cost Implications
A stream with 100,000 concurrent viewers at 3 Mbps average bitrate generates 300 Gbps sustained egress. A 2-hour live event with 1 million average concurrent viewers at 3 Mbps yields approximately 2.7 petabytes of egress, costing $54,000-$216,000 in bandwidth at CDN rates of $0.02-$0.08 per GB. Production systems use dual-path ingest (different PoPs or providers) with automatic failover when packet loss or latency exceeds thresholds, ensuring critical events continue even if one path fails.
Key Insight: The ingest/delivery separation is fundamental to streaming architecture. RTMP handles the hard problem of reliable live capture; HLS/DASH solve the scaling problem by converting live streams into cacheable HTTP segments that CDNs distribute efficiently.

💡 Key Takeaways

✓Streaming splits into ingest (broadcaster to server, RTMP/SRT/RIST) and delivery (server to viewers, HLS/DASH)

✓RTMP provides 1-3 second ingest latency over TCP; SRT/RIST use UDP with ARQ for better loss recovery

✓HLS/DASH delivery enables CDN caching and ABR streaming, working over standard HTTP ports

✓100,000 viewers at 3 Mbps generates 300 Gbps egress; 1 million viewers for 2 hours costs $54,000-$216,000 in bandwidth

📌 Interview Tips

1Explain the ingest/delivery separation: RTMP handles reliable capture, HLS/DASH enable CDN scaling

2Calculate bandwidth: concurrent viewers times average bitrate equals egress (100K at 3 Mbps = 300 Gbps)

3Mention dual-path ingest with automatic failover as standard practice for critical live events

← Back to Streaming Protocols (HLS, DASH, RTMP) Overview