Mid-stream
The switch happens inside the playing session, on a segment boundary, without a restart and without the viewer doing anything.
Solutions · Live sports
Kickoff concurrency.
Built for the night of the match, not the average Tuesday.
The spike
The audience arrives in one minute.
A match does not ramp; it detonates at kickoff. Concurrency multiplies faster than any single CDN allocates capacity, and match night is the worst possible time to discover a vendor’s limit. Spreading the load per viewer across the CDNs you already contract means no single vendor’s ceiling becomes the broadcast’s ceiling.
The rescue
The viewer never moves. The source switches.
When a CDN strains in the second half, each affected session’s next segment comes from a healthier CDN — behind the playing buffer, per viewer, mid-stream. The goal is not interrupted by the failure, and not by the rescue either. Viewers who joined before the trouble are exactly the ones a DNS answer cannot reach; they are the ones this mechanism exists for.
Per viewer
The sessions that are suffering switch. The stadium’s worth of sessions that are fine stay put. No mass migration, no secondary stampede.
Nothing to install in the venue
The mechanism is standards in the stream — CMCD up, Content Steering down — spoken by the players you already ship.
The money
Ad breaks are load-bearing.
SCTE-35 markers drive the insertions the whole rights model rides on. vidNT parses them and holds switching inside no-switch windows around each break, so a rescue never cuts across an ad avail. The switch waits for the break to end; the break never waits for the switch.
Prove it before the season does.
Run your own stream through the mechanism on a quiet Tuesday, and read the log.