Relay is an edge network built for live video: one ingest endpoint, sub-second glass-to-glass latency, and automatic origin shielding — so a spike in viewers never reaches your encoder.
Push a single RTMP or SRT stream to Relay. We transcode once, cache the ladder at the nearest edge to each viewer, and shield your origin from concurrency spikes.
SRT and RTMP ingest, LL-HLS and WebRTC egress. No re-buffering queue between hops.
Your ladder, respected end to end — Relay doesn't re-encode renditions you already built.
A concurrency spike hits the edge cache, not your encoder. One request per rendition reaches origin.
Point your encoder at one URL. Relay handles edge selection, retries, and failover between points of presence automatically.
Aggregate health per region, refreshed every few minutes.
| Region | Edges | p50 latency | Status |
|---|---|---|---|
| EU-West | 9 | 142ms | ● operational |
| EU-Central | 7 | 165ms | ● operational |
| EU-North | 5 | 171ms | ● operational |
| US-East | 8 | 188ms | ● operational |
| APAC | 6 | 203ms | ● operational |
| Capability | Relay | General CDN |
|---|---|---|
| Sub-second live latency | ✓ built in | add-on, extra cost |
| Origin request coalescing | ✓ per rendition | per edge node |
| SRT / RTMP ingest | ✓ native | often unsupported |
| Billing | flat per-channel | per-GB egress |