Net Video Downloads Test Limits

From a ZDNet article, "Net video explosion triggers traffic jam worries:"

"The amount of video online is skyrocketing, whether it's "Lost" episodes or movie trailer mash-ups. The phenomenon is putting new stress on ISP networks, which are seeing the demands on their bandwidth burgeon.

Now a new wave of companies--some newcomers, some with familiar faces--are stepping up to play the role of traffic cop, arguing that they have ways to manage this surge in video traffic and keep networks healthy.

"Everyone loses in the current scenario," said Michel Billard, a former HP executive who recently joined start-up Itiva, one of the companies offering video-speeding technology. "What we need is a way to amplify the bandwidth that's available."

Itiva's technology works by taking a huge movie file and breaking it up into tiny individual pieces that are formatted just like ordinary Web pages. When they're downloaded by a user, these individual pieces--Itiva calls them "quanta"--are stored in ISPs' Web caches, which are already distributed in every network.

Once stored separately like this, they can be quickly downloaded and pieced together by anyone else in that network, in a way that's much more inexpensive for the ISP than if everyone was going back to the original download site.

"It's like BitTorrent for ISPs," said one top network company executive familiar with the technology, but who asked not to be named. "It's very much of interest to telcos."

I have also seen a solution from NFT.

From the NFT site:

"NFT is leading the use of distributed systems to overcome this broadcast bottleneck. The basic idea behind such systems is that each viewer's computer will not only act as a receiver of a video stream, but also as a rebroadcast node - passing a copy of its stream onto a small number of other viewers (in the basic NFT system either one or two other viewers).

The major challenge faced by those pursuing the distributed broadcast approach is how to manage the resulting distributed network to provide stable and reliable service when large numbers of viewers are continuously tuning in (joining the network) and tuning out (departing the network)."