This story appeared on Network World Fusion at
http://www.nwfusion.com/columnists/2001/0611bradner.html

'Net Insider:

The missing Internet



By Scott Bradner
Network World, 06/11/01


I don't know how they could miss the Internet, but just about everyone - from pundits to venture capital companies - seems to be oblivious to the lessons of the Internet when thinking about the architecture for the future of data networking.

Three of the most basic lessons of the Internet are that you don't know what you're going to be doing, you don't know who you'll be doing it with and you don't know how many you will be doing it with at the same time.

Specifically, the Internet is all about simultaneous, application-agnostic, general connectivity to multiple sites. For example, downloading a single Web page may involve communications with three or more Web servers - the originating Web server, a content distribution server and a server providing the banner advertising. And the next click can take you to a totally different site.

But somehow all too many folks forget about the Internet whenever they talk about future network architectures.

The latest example is Tony Lavia's otherwise quite reasonable article "The Next Next-Generation Network" in the May issue of Business Communications Review. At least he mentions the Internet or at least uses the term "Internet-type" when referring to data networks, but then goes on to talk about Multi-protocol Label Switching (MPLS) and VPNs, which are just what the Internet is not.

Most articles I've seen on the topic never mention the Internet at all. I don't think Lavia meant to dismiss the Internet as the present and future general data network, but too many others seem to think the Internet does not actually work - at least in the sense that the phone network works and gives predictable service quality. (Of course one has to ignore cell phones when making any claim about phone network quality of service.)

These observers think the Internet needs to get fixed, and the way they think it should get fixed is to use per-instance of application circuits. They do this as if chanting the term "circuits" as a mantra will fix all problems.

I heard the phrase "The Internet does not work" three time in less than two hours at an advanced technology session a few months ago. Two of the three speakers wanted to fix the broken-only-in-their-minds Internet by introducing circuits - generally with MPLS.

But circuits, by definition, do not provide the type of anyone, anytime connections that the Internet has proven are needed. They fit a different Internet - one in which some service provider defines who you can talk to and when. Not nirvana to me.

There are circuits in today's Internet and there will be in tomorrow's, but they do not get close to the user - they are only in the innards of the ISPs and many of those circuits might be going away. Remember the Internet - it does work.

Disclaimer: The Harvard medical school worries about the innards of people, but has not expressed an opinion on the innards of ISPs.

All contents copyright 1995-2002 Network World, Inc. http://www.nwfusion.com