David Farber, a veteran of Internet technology and politics, dropped by Cambridge, Mass. today and was gracious enough to grant me some time in between his numerous meetings. On leave from Carnegie Mellon, Dave still intervenes in numerous policy discussions related to the Internet and “plays in Washington,” as well as hosting the popular Interesting People mailing list. This list delves into dizzying levels of detail about technological issues, but I wanted to pump him for big ideas about where the Internet is headed, topics that don’t make it to the list.
How long can the Internet last?
I’ll start with the most far-reaching prediction: that Internet protocols simply aren’t adequate for the changes in hardware and network use that will come up in a decade or so. Dave predicts that computers will be equipped with optical connections instead of pins for networking, and the volume of data transmitted will overwhelm routers, which at best have mixed optical/electrical switching. Sensor networks, smart electrical grids, and medical applications with genetic information could all increase network loads to terabits per second.
When routers evolve to handle terabit-per-second rates, packet-switching protocols will become obsolete. The speed of light is constant, so we’ll have to rethink the fundamentals of digital networking.
I tossed in the common nostrum that packet-switching was the fundamental idea behind the Internet and its key advance over earlier networks, but Dave disagreed. He said lots of activities on the Internet reproduce circuit-like behavior, such as sessions at the TCP or Web application level. So theoretically we could re-architect the underlying protocols to fit what the hardware and the applications have to offer.
But he says his generation of programmers who developed the Internet are too tired (“It’s been a tough fifteen or twenty years”) and will have to pass the baton to a new group of young software engineers who can think as boldly and originally as the inventors of the Internet. He did not endorse any of the current attempts to design a new network, though.
The full interview: http://radar.oreilly.com/2012/01/a-discussion-with-david-farber.html