Joost Has Some Infrastructure Challenges

21 Comments

With $45 million in the bank, Joost seems is all set to become a major player in the world of re-broadcasting produced content from the creative factories of large media companies. But before it does that, the company needs to overcome some technological challenges that could seriously degrade the company’s service.

At the time of Joost’s limited beta launch, the service melted down and became unavailable for many users. Despite server fixes, as recently as yesterday, tipsters were writing in to tell us the service was down again. And the underlying technologies at Joost to enable the product’s peer-to-peer streaming may make this a regular occurrence.

Joost is using a peer-to-peer version of streaming video, which means that it will have to deal with quality of service issues that come along with both P2P and streaming. Those in the technical community believe it would be difficult for the company to guarantee the quality of service of the video streams. Michael Wolf, of ABI Research on his blog, rightfully points out some of the chinks in Joost.

While Joost is not 100% P2P powered, as they have indicated they will use some servers/CDN infrastructure, they do rely on P2P in large part for content distribution. And the fact that much of their content is pretty long-tail (meaning its not exactly what most of us are hankering to watch), the chances there will be someone in close proximity to me (or in other word, a PEER in P2P) to act as a source is pretty low.

This is the weak link. If you are watching some crappy CBS program that is coming from a peer, and that peer decides to do something else with the bandwidth, there is going to be a quality hit. As Anil Gupte, in his piece “7 reasons why Joost could fail,” writes:

According to an unverified source, a Dutch ISP has tested Peer-to-Peer Streaming in the field and found that approximately 40% additional overhead is to be expected in using this technology. It is not clear whether the overhead is in the bandwidth (due to retransmission) or in processing (due to reconfiguration of the sources). However, it makes sense that there is overhead when combining buffers from multiple peers. And if one of the providers becomes unavailable, there will be tremendous overhead in reconfiguring the others.

Another big issue, at least in the near term, is the many of us have upstream bandwidth limitations; the same stuff that bogs down our use of BitTorrent. There won’t be enough bandwidth for folks to upstream shows.

In other words, Joost will have to get a lot of users in countries where there are fat upstream connections, and hope that people are consistently online watching Joost all the time. It worked for Skype, but I am not sure if it is going to work for video.

Joost, apparently is going to use some sort of a content distribution network to make up for it all, but that also means the P2P economics that worked so well in the case of Skype may not be an advantage for them at all.

For some of us who are old enough to remember Skype’s early days, the service went from a little company to one with millions of users, causing noticeable quality degradation. If that could happen in the case of voice, video brings a whole new set of complexities, and problems.

21 Comments

Steve Schear

This problem of assuring long-tail content availability is part of what we tried to solve at Mojo Nation (the venture where Bram Cohen learned many of his P2P tricks for Bit Torrent). Our solution, provide a market for pricing P2P resources (e.g., bandwidth, disc storage, etc.) so that real incentives can be offered for clients to cache less popular content.

Yaron

These are excellent points Om. The Internet will not withstand the surge of video, TV and film content being thrust onto IP pipes without P2P – BUT — P2P video distribution is ideally suited for complimenting traditional CDN servers, not replacing them. When seamlessly offloading peak/burst bandwidth from central servers, p2p shines. When trying to deliver across the entire long tail of content, pure p2p is often overkill and unreliable. The Estonians will innovate on this front no doubt, but they will of course rely heavily on central “super-seeds”.

Pando Networks, a hybrid p2p media distribution platform, has spent over 2 years optimizing p2p efficiency for downloadable HD media in both one-to-many and one-to-few networking scenarios, and let me tell you, its not easy. For streaming it’s even harder to get right because the slightest performance hit is immediately visible to consumers. You have to balance three things simultaneously; bandwidth efficiency, bitrate performance and perhaps most importantly (and least talked about) minimizing the use of the consumer’s PC resources so that they don’t “feel” your p2p app. As an old push technology guy I can’t stress this enough for upcoming p2p video startups.

Comments are closed.