NewTeeVee has a load of questions and a poll going on about the evolution of video distribution on the Internet.

Personally I vote for hybrid networks, i.e. P2P+CDN, when it comes to downloading large files (video or otherwise) across the Internet. For me this applies to publishing large files (one to many) or exchanging large files between individuals (one to few ~ e-mail). That is what we have been implementing for Podmailing.

To my mind there are obvious reasons to avoid a pure-P2P or a pure-CDN system from a business perspective:

• CDN bandwidth costs are fully covered by you while P2P bandwidth doesn't cost you a dime
• You can't have complete control of a P2P network but you do have total control of your CDN - that's a major issue to maintain a high quality of service
• P2P is great to manage peak demand for hot new content but does not guarantee anything when it comes to long tail content
• CDNs are costly when used for peak demands, but they can also guarantee quality for long tail content

Manage a smart hybrid integration between P2P and a CDN and you can get the best of both worlds:

• Reduced bandwidth costs, off-loaded to non-peaks rates which are way lower
• High efficiency during connection peaks
• Guaranteed quality of service both for high-demand and long-tail content

Now, if you really need streaming, it appears that a CDN is the only viable way to go. Joost is trying to challenge that with a hybrid streaming-P2P-CDN system, but the math of scaling up is not helping them at all as noted on NewTeeVee's earlier post.

As I first noted here, and got confirmation here, they are only saving a small part of the bandwidth cost (up to a third), so their advertising dollars need to stack up very high if they want to get any close to profitability. I do salute the effort and the innovation of Joost though. I am just convinced that streaming-P2P volumes will remain insignificant compared to P2P-downloads for very long.