I hate to blow my own trumpet, but I am pretty sure I coined the phrase that the consumer ideal is “the content they want to watch, where they want to watch it, when they want to watch it”.
I certainly wrote it a lot 15 or more years ago, and it seemed to stick. I’m sure you’ll tell me if you think someone else can lay claim to it.
When I started writing it, though, the problem was that it was not practical. Asking for a programme and it starting at your convenience was impossible, short of recording it first on a VCR. There was no bandwidth around to carry individual programmes. We had to wait until digital cable, and of course now widely available broadband and 4G/LTE mobile.
It all works now, of course, and we have all enjoyed its benefits. But delivering it is not quite as simple as you think, nor is it going to get any easier. This may surprise you, but if anything the problems are growing.
In theory, if you ask, say, BBC iPlayer to start, say, episode 9 of Doctor Who, then the BBC’s server goes away, finds the right version of the file (screen resolution of your device, codec, transport layer and so on) and sends it to you. You need an internet path from the BBC to you. If your neighbour also wants to watch the same episode, even in the unlikely event of her pressing start at exactly the same instance, she will need a completely separate internet path.
This is clearly massively inefficient. There is a wonderful man called Josh Stinehour, who is a business analyst who works for the broadcast specialist Devoncroft Partners. If you ever get a chance to hear him speak, go for it. You will be exhausted at the end of the presentation, but you will have learnt a lot and laughed more.
He calculated, because it is the sort of thing that business analysts do, that to convert all television watching in the USA to online streaming and video on demand would need six new nuclear power stations to power all the additional circuits.
So the video on demand providers use a couple of tricks. First, they do not store all the content in all of the formats, because that would need massive servers. They store each programme in a house format, then transcode it and wrap it with the right metadata at the moment you request it. They talk about “just in time packaging”. It also allows them to add timely advertising into the stream without you noticing.
Then they use a content delivery network, or CDN. These specialist businesses – Akamai is a famous name, but there are many others – minimise the amount of trunk connections you need by having multiple server sites dotted around the geography that you are addressing. These are called edge servers.
Imagine that 100 people in Cardiff request a programme from an on-demand service provider based in London. Without a CDN, 100 connections have to go from the provider to a hub in London, then 100 connections go down the trunk route to the Cardiff hub, where they are broken out into 100 individual feeds. With a CDN, you transfer the content from London to Cardiff once, then distribute it from there.
The good CDNs work with content providers to optimise the placement of the content. If they know that a movie will be hugely popular in Paris but not in Prague, they might place 10 copies in the Paris edge server and only one in Prague. Or maybe not even store it in Prague but feed it individually if the requests really are rare.
Minimising the load on the internet is going to be increasingly important. According to Cisco’s latest fortunes, by 2021 the annual global traffic on the internet will more than three zettabytes. That is 3 followed by 21 zeroes. And 80% of all internet traffic will be video.
We will continue to demand the content we want, when we want it, where we want it, and on our preferred device. But it will be a challenge to make that happen.
 |
Guest blog by:
|
