alex.moskalyuk

The answer is, naturally, RIAA and MPAA, as Van Jacobson’s lecture at Google suggests. These organizations and their lawsuits have led to the quick improvements to the peer-to-peer mechanisms, introduced by the original Napster, and currently involving at the BitTorrent level. Van Jacobson regrets the falling levels of networking research that he observes right now, and points that the protocols that govern current Internet are not coping with the current congestion problems we might experience.

Not that the protocols are bad, it’s just that the problem has changed. Previously a lot of networking research went into delivering data from point A to point B, and from point B to point A. TCP/IP therefore perfectly accommodates two hosts on the Internet talking to one another. However, modern data consumption trends indicate that consumption is rarely happening between two equal hosts. More often than not a single host (CNN, NBC, YouTube) is serving requests from hundreds of thousands of other hosts (users), with no ways to optimize the delivery of the same content on the protocol level.

Van Jacobson also doesn’t like the idea of binary connectivity. You’re either on the Internet or you’re off. There’s no way to be on the network, retrieve the information from hosts and caches close to you, and be denied from retrieving information from further hosts, until you get a full connection. Overall, a pretty good lecture with lots of questions that nobody else seems to be asking, primarily because most of us are used to the current ways networking works, and don’t have any inspiration to change the status quo.