Actually, P2P does work. It works for short form, low quality downloadable (or even progressive download) content.
It doesn't work for a TV like experience. The reason is quite simple. Domestic broadband is asynchronous. That is, you might get 2Mbps downstream, but only 256Kbps upstream. So, if you are receiving a 1Mbps stream (the average global connection to Narrowstep's streams currently stand at 1.2Mbps), you can return a maximum of 200Kbps to the next user. Now, if there are ten thousand people watching the same stream at the same time this has some utility.
I calculate that it will save around 25% of network capacity, so it still means that you need a serious unicast (streaming from the server) network.
So, in the image on the left, A. is receiving a 400Kbps file, whilst, ast best, she is pushing out three 75Kpbs streams through a1, a2 and a3.
In the image below, A. is now receiving 1Mbps, but can still only return a maximum of 75Kbps (and doing this reliably with a stream is very much more difficult than with a download).
However, if the content is
being narrowcast from a VoD service this situation is exasebated - the fewer people accessing a file, the less scope for using P2P technology.
Of course, P2P companies have integrated a number of work arounds, including rolloing over onto streaming servers or using 'passive', ie always on, connections so that the broadband connection of people not using the service or accessing the same clip can be utilised, effectively implementing a covert grid network (this is how Joost works).
This results in a further problem. Content owners and advertisers worry about quality of delivery (this is one of the main reasons IMHO that you haven't seen much advertising on the UGC sites), so cutting costs on delivery compromises revenue generation in a serious way.
P2P has its place, but its usefulness is limited in a commercial online broadcast context.
I fail to see why so many online TV companies are tilting at this windmill.
It doesn't work for a TV like experience. The reason is quite simple. Domestic broadband is asynchronous. That is, you might get 2Mbps downstream, but only 256Kbps upstream. So, if you are receiving a 1Mbps stream (the average global connection to Narrowstep's streams currently stand at 1.2Mbps), you can return a maximum of 200Kbps to the next user. Now, if there are ten thousand people watching the same stream at the same time this has some utility.
I calculate that it will save around 25% of network capacity, so it still means that you need a serious unicast (streaming from the server) network.So, in the image on the left, A. is receiving a 400Kbps file, whilst, ast best, she is pushing out three 75Kpbs streams through a1, a2 and a3.
In the image below, A. is now receiving 1Mbps, but can still only return a maximum of 75Kbps (and doing this reliably with a stream is very much more difficult than with a download).
However, if the content is
being narrowcast from a VoD service this situation is exasebated - the fewer people accessing a file, the less scope for using P2P technology.Of course, P2P companies have integrated a number of work arounds, including rolloing over onto streaming servers or using 'passive', ie always on, connections so that the broadband connection of people not using the service or accessing the same clip can be utilised, effectively implementing a covert grid network (this is how Joost works).
This results in a further problem. Content owners and advertisers worry about quality of delivery (this is one of the main reasons IMHO that you haven't seen much advertising on the UGC sites), so cutting costs on delivery compromises revenue generation in a serious way.
P2P has its place, but its usefulness is limited in a commercial online broadcast context.
I fail to see why so many online TV companies are tilting at this windmill.
Comments