Broadcast Economics on a Managed Network

Streaming over the open internet sends every viewer their own copy, which is the only thing that can work when your audience is scattered across networks you do not control. It also means a channel that everybody wants to watch at once is the most expensive thing you operate, and that the moment of maximum audience is the moment of maximum risk. Broadcast never had that problem, because a transmitter sends once regardless of how many receivers are listening. Multicast brings that property onto an IP network, where a single stream is carried through the network once and delivered to every receiver that has joined it. On a network you control, the cost of a channel becomes largely independent of how many people are watching it.

It is worth being direct about the boundary, because multicast is frequently asked about by people it cannot help. If your viewers reach you over the public internet on connections you have no relationship with, multicast is not available to you and unicast delivery from an edge network is the correct answer. If you are a telecoms operator, a cable company or a hotel, campus or building network delivering to set top boxes across infrastructure you run, then it applies and the savings on popular channels are substantial. Most Vodlix deployments are the first kind, which is why delivery is normally discussed in terms of the global CDN. Operators in the second category are a distinct case, and it is worth saying so plainly rather than implying everybody gets broadcast economics.

Multicast is efficient precisely because it is one stream, and that has consequences. Everybody joined to it is at the same point, so the personal features viewers now expect on demand content become harder or impossible on the multicast path itself. Pausing, rewinding and restarting a programme are per viewer behaviours, and delivering them alongside multicast usually means a second unicast route for the viewers who ask for them. That is a normal architecture rather than a failure, but it should be planned rather than discovered. Time shifting is covered under time shift and recording under network DVR, both of which are unicast concerns even where the base channel is not.

The realistic shape for an operator with a managed network is multicast to the set top boxes on that network and unicast to everybody else, which means the mobile apps, the browser and any subscriber outside the footprint. That is not a compromise so much as the correct answer, since the two delivery methods suit different situations and one platform can carry both. What matters is that the catalogue, the entitlements, the guide and the subscriber records are shared regardless of how a given stream reached somebody, so a viewer moving from their set top box to their phone is the same customer. The channel experience itself is described under live TV and the wider operator context under IPTV.