Last month, Bram Cohen introduced Bit Torrent Live, a strategy that allows the live streaming of video in a similar way to the way Bit Torrent’s “traditional” service.
Like Bit Torrent ‘free’ or ‘plus,’ Bit Torrent live holds that as your numbers increase, scaling occurs as peers share the traffic, and reduce bandwidth requirements.
In an interview with Torrent Freak, Bram talked about the way the system works:
“Doing live streaming well on the Internet has long been a problem. Peer-to-peer live streaming has always suffered from high latency, meaning there is typically a lot of delay between when a broadcast happens and when end users see it, typically dozens of seconds or minutes,” Cohen told TorrentFreak.
“BitTorrent Live allows a broadcaster to stream to millions of people with just a few seconds of latency. This is new, and unique, and potentially world-changing.”
Bram Cohen explains that the patent is in no way going to restrict user access to the new protocol. Quite the contrary. BitTorrent Live will be available to end users for free, and publishers who are using the service and hosting it on their own will not be charged either.
“We want people to use and adopt BitTorrent Live. But we aren’t planning on encouraging alternative implementation because it’s a tricky protocol to implement and poorly behaved peers can impact everyone. We want to ensure a quality experience for all and this is the best approach for us to take,” Cohen told TorrentFreak.
BitTorrent Live is a complex technology, but basically works by dividing peers into various “clubs” of peers who share data among each other via a UDP screamer protocol.
“To get slightly more technical, the way BitTorrent Live works is by making subsets of peers responsible for subsets of data. High robustness and low latency is achieved by using a screamer protocol between those peers,” he explained.
“For the last hop it uses a non-screamer protocol to regain congestion control and efficiency. There is redundancy and some waste in the screaming, but that’s kept under control by only using it to get data to a small fraction of the peers.”
Now here’s the question: Do we need mix streams or not? Tsahi Levitt-Levi did an excellent job talking about traditional multipoint conferencing on his blog, but the question has to be asked: Is that the future?
If you look at what we’re seeing on TV today, we have shows like CNBC and CNN putting several people on the screen at once and not mixing the streams. Gathering all the streams would in effect make the human speakers figure out the protocol of prioritization, rather embedding in the system.
So if we have a bit torrent live conferencing system and we don’t include an MCU, will it be compelling?
BitTorrent Live with WebRTC would allow us to see a convergence of lightweight compression with transport beyond what we have today.
Like many other Innovator dilemmas (thank you, Clayton Christensen) this strategy will look inferior for some time. I have lots of friends who turn up their noses at Google Hangouts, but as we see more streaming strategies come of age, commentary and training services might be the early adopters of this strategy.
Adoption then will depend on how compelling the client software is in managing the streams.
If I were looking for the blue ocean of WebRTC, this is where I would focus.
Edited by
Braden Becker
