Been There, Didn't Do That

By: Scott Bradner

Coming back from two weeks in Europe I almost thought I had gone through some sort of time warp. After I got back I started to catch up on the trade press from when I was away and from the few weeks prior to that when I was a bit over-busy to keep up. I kept seeing headlines that I could have seen five years ago.

Headlines proclaimed that "ISDN does it all", that X.400 and X.500 were just around the corner, and client/server was all the rage. For a while I was quite confused but then I saw a few headlines about the wonder of ATM and I realized that what I was experiencing was remembrances of futures from the past.

The whole thing got me to thinking, a bit fuzzy from jet lag, but thinking none the less. What was it that kept the headline writers and, as a bit of reading showed, the reporters as well, believing, year after year, in the same sort of line from the vendors.

Partly, I guess because the same reporters have said the same things every now and then for many years and would like to be right at some point. Sort of like me and my stubborn holding to the belief that Apple Computer will actually do something (anything) to the actual benefit of us long-term Macintosh supporters.

Another factor is the attractiveness of the promise. ISDN has been touted as the total solution to all of personkind's data networking problems. Even if some ISDN proponents are now willing to admit that there are other useful networking technologies, they still claim that ISDN is the way to get cheap data networking to the masses. A claim that, so far, has wilted in the face of the tariffs that most of the local telephone companies have filed.

X.400 holds out the promise of universal electronic exchange of an endless variety of data including text, voice and video. X.500 promises a universal directory, you can look up anyone, anywhere, and get their electronic address.

The meaning of client/server has changed over the years from small machine talking to a mainframe to PC talking to a bigger PC but the claim has remained the same: more performance and flexability for less cost. Another attractive promise.

There is a common thread running through these ideas that may not be apparent at first glance. All of these technologies, in one way or another, make the assumption that someone is providing the network and part of the network is centralized servers. This is particularly true with X.400 and X.500. These protocols were designed with the notion that the phone companies (PTTs in Europe and elsewhere) would be providing the network to all the users and that they would have directory servers and email forwarding agents as part of the network service offering. In this picture, the network is more than a connection between users. It also includes quite a bit of hardware to provide services that the network provider assumes the users want. This is what Einar Stefferud refers to as "complexity in the core". This same model is what Prodigy and Compuserve started with.

A problem with assumptions is that they do not always match real well with the real world. The most successful of today's networks (the Internet) does not share these assumptions. The Internet basically adheres to the model that everyone can be a server, that function is not restricted to a few network-provider-provided servers. Prodigy and Compuserve have figured this out and now provide open access to the Internet with its lack of central control.

It may be that the headline writers are correct and the Internet model is on its way out, but I kinda don't think so. In fact when you take a close look, many of the X.400 and X.500 products no longer assume servers in the network; they use X.400 or 500 internally and speak Internet mail to the rest of the world.

disclaimer: Historically Harvard has demonstrated a disbelief in central services (and control), but empirical evidence led me to these opinions.