In the future, the dominant traffic on the internet will be video. However, it will not look like TV. Instead it will be more like video on demand, for everybody.
The early internet was mainly used for interactive terminal traffic, but that soon gave way to file transfer. In the late nineties, web traffic took over. In the past year, peer to peer (i.e. file sharing) has become dominant, within which the major traffic type is probably audio and video.
When I asked the CEO of a large European Internet exchange what his take was on bandwidth usage in the future, he expected it to be 100 Mbit/sec to every home, in order to accomodate video feeds. “That is what I have”, he smiled.
A senior Cisco executive recently pointed out to me that BBC’s iplayer, which distributes BBC content in the UK, has taken only half a year to grow to 10% of traffic on British ISP’s. And that is just the beginning. In addition, videoconfercing has replaced 80% of his travel.
For digital infrastructure architects there are important subcategories in this traffic. There is streaming traffic, which goes from server to the viewing device (PC or otherwise) without any intermediate storage, and there is nonstreaming traffic that is first stored with the user. I expect the balance to tip in the direction of streaming traffic. This is more in line with the websurfing and channel surfing type of behaviour that is the result from an abundance of information.
For similar reasons I do not expect multicast or broadcast models to become dominant. These are more traditional models in which a group of viewers receives the video feed simultaneously. The reality is that there is a very limited interest in realtime video. The major types I see are sports, videoconferencing and security. Paradoxically, only a fraction of news will have a realtime audience.
So, there is every reason to expect video feeds to dominate Internet traffic. The question is how many of these feeds will be from security cameras.