Live Production as Infrastructure
What African media teams get wrong about streaming — and the boring engineering that wins on Sunday morning.
African media teams — churches, conferences, broadcasters — tend to think about live production as an event. A Sunday service, a Friday night summit, a one-off keynote. Hire the camera operators, rent the switcher, push to YouTube, breathe out. The teams that scale think about it differently. They think about live production as infrastructure: a repeatable, measurable system that has to perform every week, with the same uptime expectations as a payments platform.
The first re-frame is the encoder. Most African production setups still rely on a single laptop running OBS, plugged into a single ISP, streaming to a single destination. That is three single points of failure stacked on top of each other. The infrastructure-grade answer is a hardware encoder (Teradek, Pearl, LiveU) or at minimum a dedicated streaming PC, dual-bonded internet (one fibre, one 4G/5G as failover via a Peplink or LiveU bonder), and a primary-plus-backup ingest configured on the streaming platform. The cost difference is real but bounded. The reliability difference is the difference between 'we went off air for forty minutes' and nobody noticing.
The second re-frame is the CDN. Pushing directly to YouTube Live is fine for a single distribution surface. The moment you also need to feed Facebook, an app, a partner broadcaster and a backstage monitor, you need an intermediate layer — a restream service like Castr or, at scale, a self-hosted SRT relay into a multi-CDN setup (Cloudflare Stream plus a fallback). This separates 'capturing the show' from 'distributing the show', which means a problem on one CDN does not take down the whole broadcast.
The third re-frame is observability. Treat the stream like a production service. Synthetic monitors that pull the HLS manifest every thirty seconds. A simple Grafana board for bitrate, dropped frames and viewer count. An on-call rotation for the broadcast window with a runbook. None of this is glamorous. All of it is what separates a media operation that can be trusted with a corporate sponsor from one that cannot.
And finally, the run-of-show. The single highest-leverage document in live production is not the rundown — it is the technical run-of-show: every cue, every camera change, every lower-third, with the named human responsible and the named fallback. Print it. Tape it to the desk. Walk through it ninety minutes before doors. The shows that go smoothly are the shows where this document exists. The shows that go wrong are the shows where everyone 'knew what to do'.