<b>The streaming site that went dark when its CDN had a bad day</b>

<b>The streaming site that went dark when its CDN had a bad day</b>

A live-events streaming startup ran on a single CDN. One afternoon that provider had a regional outage in Europe; their entire EU audience — 30,000 concurrent viewers — got connection errors for 50 minutes. One vendor's bad day was the whole company's outage.

We added a second CDN and put DNS-level health-checked steering in front of both. Traffic now defaults to the primary, but if a region starts returning errors or latency spikes, the resolver shifts that region to the secondary within seconds.

The next time the primary had a London hiccup, EU viewers were automatically re-steered to the backup CDN. Measured viewer-facing error rate during the incident stayed under 0.3% instead of hitting 100% in-region. No manual intervention, no war room.

The number that mattered: in-region errors capped at 0.3% during a full CDN outage.