Broadcasters buying connectivity are frequently sold on a headline number, and frequently disappointed by it. The reason is that broadcast traffic has almost none of the characteristics that ordinary internet products are optimised for, and most of the ones they are not.

Four ways broadcast traffic is unusual

It is sustained. A live feed runs at a constant rate for its entire duration. There are no quiet moments for a contended service to recover in. A circuit that averages well over a day can still fail an hour of continuous output.

It is upstream. Contribution — getting material from a location back to the studio — runs in the direction most consumer-derived products treat as an afterthought. Asymmetric services are sized for download.

It is scheduled. Demand is not random. It is concentrated precisely when the audience is largest, which is precisely when everything else on a shared network is also busiest.

It fails publicly. A slow file transfer is an inconvenience. Dead air is the failure everybody watching notices at once, and the advertiser notices immediately afterwards.

Taken together these mean a broadcaster needs committed capacity — bandwidth reserved for them whether or not they are using it — rather than a contended service with a high ceiling. It costs more per megabit, for the straightforward reason that the provider cannot sell the same capacity twice.

The three paths, and buying only one

Most broadcasters procure one of these and meet the other two during an incident.

Contribution

Material from a stadium, a hall, a regional studio or a field location has to reach the origin point with enough headroom that the encoder is making quality decisions rather than survival ones. Where there is fibre, fibre. Where there is not — which is most locations that generate news — satellite, sized for sustained upstream rather than headline download.

Distribution

From origin out to transmitters, platforms and audiences. On fibre this means committed capacity on protected paths. For the parts of the country fibre has not reached, satellite carries the same feed, which is a considerably older and better-understood use of satellite than internet access is.

The backup for both

Rarely bought, invariably wanted. The useful property of satellite as a protection path for a fibre-delivered feed is that it shares nothing with it — no duct, no route, no landing station. Two fibre circuits into the same building frequently share all three.

Origin, caching and the ocean problem

There is a cost most broadcasters carry without examining. If content is hosted outside the country, every viewer request crosses a submarine cable. The same file travels the same route once per viewer, and the broadcaster or their platform pays for that capacity every time.

Holding content in country — origin, caching, archive — changes that arithmetic. The file crosses the ocean once. Viewers are served from a rack in Mogadishu, which is faster for them and cheaper for whoever is paying for international capacity.

The archive point is the underrated one. Media archives are large, they grow, and they are the least replaceable asset a broadcaster owns. Holding them in a facility you can physically reach, under your own jurisdiction, is worth more than it sounds like it should be — and it costs less than most broadcasters assume, because archive storage does not need the performance characteristics production does.

What to specify

Committed rather than contended capacity, stated in the contract. Upstream sized for your actual contribution bitrate, not derived from a download figure. A protection path that is independent in kind. And, if any material is hosted abroad purely by historical accident, an honest look at what bringing origin in country would do to your transit bill.

None of that is exotic. It is just a different specification from the one a standard enterprise product is built to, and broadcasters who write it down get a service that survives the evening.