Is It Soup Yet?

By: Scott Bradner

Back in the November 8th issue one of my partners in prose on these pages talked about Commercializing the Internet. Well, doing the very sort of thing that editors love, I'd like to look at some of the same issues from a bit of a different viewpoint.

I've been associated with an Internet service provider for quite a while now. Since its founding in 1989, NEARnet has grown from 9 to about 290 members. It has gone from a pure NSF Appropriate Use Policy (AUP) compliant membership to one where 1/3 of the new members choose the extra cost option of open commercial connectivity. (Pardon me for a moment while I go pick up a horn to blow.... back again now.) Since NEARnet's founding I've been the chair of the NEARnet Technical Committee and a member of the NEARnet Steering Committee. In other words I've been there helping to build the highway. So I'm going to have a somewhat different perspective on the subject than some others might. (No, I'm not saying that Ed does not know from where he speaks, however the view, priorities and opinions are often strongly influenced by the location of the pulpit.)

First of all, we should run through what the Internet is since it might not be clear from most of the press stories. Historically the most accepted formal definition of the Internet has been "the collection of interconnected data networks that run the TCP/IP protocol". "Interconnected" means functionally interconnected in such a way that supports direct virtual connections between computers on various individual networks. Although this definition is getting a bit fuzzy with the proliferation of gateways between TCP/IP networks and others running protocols like Novell's IPX this is still the best way to think of the Internet. Note that this means you can have an Internet email address and yet not actually be on the Internet if the mail gets forwarded through a gateway of some sort.

The component networks of the Internet range from the LAN on your desk to the nation-spanning NSFnet. Some parts of the Internet have been supported during some parts of their existence with governmental funds, but the vast majority of the Internet has not been. For example, the LANs within Apple computer and Harvard University were installed and operated with no specific federal support. Likewise Alternet and PSINet were started and built with private funds.

As I mentioned in a past column, much of the actual Internet connectivity is being provided by the regional networks. Many of these were started with funds from NSF but most have weaned themselves away and are now operating on their own (NEARnet has been self-sufficient since the beginning.) These components can be called privatized now. The main part of the Internet that is not fully privatized is the inter-regional connectivity. Much of this currently flows over the NSFnet but an increasing number of regionals (including NEARnet) and all of the alternative providers like Sprint, Alternet, PSINet and ANS have arranged for alternative interconnections through the CIX. Even the residual NSFnet connection will end next year when NSF gets out of the backbone business.

The last estimate I saw indicated that in the U.S. the federal government currently pays for less than 10% of the inter-organizational network traffic. The remainder is carried over privatized networks, offering full-function reliable commercial services, unencumbered by the NSF AUP, that are paid for by the subscribers. (Harvard pays NEARnet for its connection just as Digital and Thinking Machines do.)

The Internet is over 90% privatized now (if you count the investment of the local institution or corporation this goes to better than 99.99%). There is little to be gained in asking for additional conversion since it will happen next year anyway. What we do have to focus in on is the range and cost of the Internet services and the attention to the quality of service provided. The competitive market place will drive these issues. There are now at least five Internet service providers in the New England area. In order to be successful NEARnet must focus on bringing the highest quality service it can for the best price it can, since the rest of them are doing the same thing.

The disclaimer of the month is from Lennart Regebro: Any Opinion expressed above is from (c) Rent-An-Opinion(tm). It is not an opinion of Harvard University.

sob@harvard.edu