Thursday, August 23, 2007

Is the Internet Over?


It's old, vulnerable, and overloaded. Yeah, the Net has its problems, but the thing is, it works.

By John C. Dvorak

I used to joke around about shutting down the Internet so that its protocols and basic architecture could be rewritten from scratch. I was semiserious. More recently, Elton John, who apparently can't use a computer, said the Net should be shut down for five years so that the arts can flourish. Okay, whatever. Myself and Elton John aside, we're actually now seeing serious initiatives that may result in the closing of the Internet as we know it.

There have always been undercurrents that have tried to eat away at the foundations of the current Internet. One is Internet2, a parallel-universe Internet that would be used by academia and perhaps the military. It would have ultra-high-speed file transfer without a lot of the latency issues that we experience with the current model.

A few years ago, people were chatting up Internet2, but most of that chit-chat has died down. Begun in 1996 to much hoopla, the project seems somewhat bogged down by the academicians it aims to serve. In the meantime, another parallel project, called National LambdaRail, appeared. It promoted more new technologies and ideas to achieve ultra-high-speed international networking. It recently merged with Internet2.


Meanwhile, our old-fashioned plain-vanilla Internet is seriously planning on changing from IP version 4 to IP version 6. This is, for the most part, because of the never-ending complaint that "we're running out of IP addresses." IPv6 is supposed to be able to solve some security and spam issues, too. The problem here seems to be integrating IPv6 into the current network without causing all sorts of routing complications and other problems. From what I can tell, it's a nightmare.

Japan has just announced a 7.8 billion yen project to develop all-new security-centric architectures to replace the Internet in that country. This is supposed to be rolled out by 2020. Every so often, the Japanese get an itch to leapfrog everyone, and the results have been spotty. Their last overhyped project was the 1982 "Fifth Generation" scheme whereby Japan Inc. was going to jump past all current computer technology and develop a massively parallel architecture that actually works. It generated a lot of fretting and little else.

Not to be left out are the English. The House of Lords recently demanded that the Internet be rewritten from scratch because of the "wild West" nature of the current system. The primary concern is that of identity. The "nobody knows you're a dog when you're on the Internet" kind of thing seems to upset the upper crust in the U.K., although I doubt that many of them can even send an e-mail. Still, they are now all experts.

The fact of the matter is that the Net, as designed, is more robust and versatile than anyone imagined, and the likelihood that a new Internet would be as reliable might be sheer folly. Despite predictions that the Net was overtaxed and would collapse under its own weight, it keeps humming away.

There are four current concerns about the Internet. The main one seems to be security. How vulnerable are the Net and its users to various security attacks? This involves the anonymity factor as well. Old-line thinking does not like the idea that you can hide on the Net. They fear something bad will come from this.

The second concern is that we are out of IP addresses (as mentioned above), and we've been forced to share them.

The next serious concern is the eventuality of IPTV and the likelihood that most TV will be running over the Internet someday. This has already been predicted to triple the load on the Net from the outset, with continued increases in bandwidth demand.

Then there is simply the Net's age. It's old. We know a lot more about protocols than we did in 1969 when the proto-Internet first appeared as Arpanet. Over time, the Net has been a transport mechanism for an amazing hodgepodge of protocols and subsystems, many of them quite old. Some think that newer is always better and that we need to scrap the old and reinvent the wheel.

Reinventing the wheel is problematic when that wheel is attached to a wagon that's moving. There is more to this than merely upgrading a recent "build" of AOL. The only way any change will work is literally to roll out a parallel system that can be used jointly with the current Internet. You know, like having a Mac and a PC on the same desk, or something like that.

That way we can make the change at our own pace—if we even want to change, that is. I'm not holding my breath that anything will happen for years to come.