Many of you will be reading this as you prepare for IBC. Some may even be reading it on the train/plane/ship to Amsterdam. I want to take a couple of minutes of your time to look at a last minute addition to IBC. It concerns IP, but trust me: this is good news so please do not stop reading just yet.
For very, very many years we have connected video over co-axial cables with BNC plugs on each end. Once upon a time it was analogue. Sometimes we even had to have a bundle of four identical cables for RGBS (red, green, blue and spare as a friend of mine used to call it).
When video moved to digital there was a short period when signals were sent in parallel and we had to mess about with multi-pin connectors, but very quickly someone came up with the idea of converting the digital video signal to a serial stream. That very clever person made two more brilliant decisions: first, to call video in digital form delivered serially the serial digital interface; and second to use co-ax cable and BNC connectors. We didn't even need to throw away our patch cords!
SDI became accepted as the global standard more or less instantly. We had total confidence that we could connect any output to any input with a standard BNC cable and it would work.
Now, for all sorts of very good reasons, we are moving away from SDI and moving to IP connectivity, as is used in any other computer application. And we have the same advantage: everyone understands about RJ45 connectors on twisted pair cable to carry ethernet streams.
But just because we can pick up any cable and plug it in does not mean that we have universal plug and play, the way we did with SDI. In fact we absolutely definitely do not have plug and play, because we have moved from a simple, dedicated, realtime interface standard to the need to piggy-back broadcast standard stuff onto IT transport layers.
30 years ago I spent a lot of time working with what was then called the "OSI seven-layer model" for computer-to-computer interfacing. Back then there were days at a time when I could remember what the seven layers were. But critically, the model abstracts the transport layer from the application layer. In other words, devices at both ends of the ethernet cable understand the ones and zeros the other is saying, but only know what to do with them if the right application is in place.
What broadcast needs, therefore, is some standardisation of the top layer of the OSI model, the application layer. Only then will, say, a production switcher understand the data stream that a character generator is sending it.