ARPANet Co-Founder Predicts An Internet Crisis 152
The Insultant writes "Dr. Larry Roberts, co-founder of the ARPANET and inventor of packet switching, predicts the Internet is headed for a major crisis in an article published on the Internet Evolution web site today. Internet traffic is now growing much more quickly than the rate at which router cost is decreasing, Roberts says. At current growth levels, the cost of deploying Internet capacity to handle new services like social networking, gaming, video, VOIP, and digital entertainment will double every three years, he predicts, creating an economic crisis. Of course, Roberts has an agenda. He's now CEO of Anagran Inc., which makes a technology called flow-based routing that, Roberts claims, will solve all of the world's routing problems in one go."
We've heard this before... (Score:5, Informative)
Packet switching was invented by Paul Baran (Score:5, Informative)
Re:Packet switching was invented by Paul Baran (Score:2, Informative)
Similar. (Score:3, Informative)
Flow-based routing attempts to identify flows of packets - TCP connections, related streams of UDP packets, etc. - and cache information about them. Then when future packets of the flow arrive and are successfully identified they can be handled using the cached information, rather than performing a full lookup of routing, QoS labeling, permission checking, etc.
It may also attempt to identify more things about it - such as what kind of traffic it is. Then it can do other things: Identify what quality of service it requires. For instance:
Streams such as VoIP and broadcast audio and video need low latencey and jitter (variability of transit time), but packets that are already delayed too much should be discarded, and the bandwidth SHOULD be limited. Meanwhile file transfers prefer delaying the packets to losing them but they're happy to take all the bandwidth left over from more critical stuff. So streams may go to the front of the line if they're timely but the trashcan if they're already delayed or there are too many of them than there should be.
Once a flow has been identified the knowledge can also be used to do other things with it: Find a better route that it has rights to use, give it preference if the customer's contract guarantees delivery (i.e. "you get 4 VoIP lines worth of bandwidth with high quality of service before your VoIP packets start getting best-effort handling".), perform "deep content inspection" (such as running email through a spam filter as a service), etc.
Circuit switching explicitly reserves resources through the network switches at the start of a session ("setup") and releases them at the end ("teardown"). Flow-based routing attempts to identify the flows of sessions on-the-fly, to speed routing decisions and be "smarter" about the flow - sometimes to the point of being able to emulate circuit-switched quality of service. But it doesn't REQUIRE setup/teardown and end-to-end cooperation to get things to happen. Instead, anything it can't identify goes through in the old way, with the router thinking about each packet of a flow from scratch, just as if the flow-related features didn't exist.
And Anagram is far from the only company working on it. B-) It's a major industry buzzword, on its way to becoming a (set of) required check-boxes for getting networking companies to buy your boxes.