Posted by: Tom Nolle
flow switching, Internet architecture, Optical Networking
An MIT researcher says that the Internet can be made 100 times faster and cheaper at the same time, and while the claim is correct in one sense, it’s a gross exaggeration in another. The idea is that optical “flow switching” could speed the handling of packets (increasing speed) and at the same time reduce the cost of devices. That would speed the Internet and make it cheaper, right?
It’s right as long as you’re talking about the Internet as a core network. It’s not at all clear that flow switching would have much impact on the edge of the network where users connect, and thus that it would improve subjective performance. Furthermore, the cost of core routers is such a small part of total Internet service cost that savings there might not even be noticed.
The technology is interesting, and it will likely change online performance and cost over the long term, but don’t look for any changes in the next couple of years at a minimum. It’s also interesting to note that the story has a kind of counterpoint in a comment by a computing technology site that computer hardware and chip architectures aren’t covered like they used to be because they’re not where the action is any more.
That’s exactly where networking is heading. The service layer and middleware are where profits and success must now be built.