The following text is copyright 1998 by Network World, permission is hearby given for reproduction, as long as attribution is given and this notice is included.

Is there hope for DSL?

By Scott Bradner

A feature of being a columnist in a technical publication is that I am presented with a surfeit of opportunities to speak to marketing droids from companies which, according to the spiel, have the answer to questions I did not know I had. (A disclosure - I do try to arrange some of the visits for lunch time so I can at least get a free meal out of the encounter.) From time to time one of these encounters results in useful information. One of those occasions came last week when Rick Gilbert of Copper Mountain came by (unfortunately the only time that worked was early morning so no free lunch.)

I've been hearing horror stories for the last few years about the problems encountered when ISPs and other companies attempt to deploy various types of digital subscriber line (DSL) links to their customers. Particular problems mentioned have included the difficulty in getting wires from the phone company which are clean enough to provide good performance, potentially severe crosstalk which results from having more than one DSL link in the same cable bundle running down the street, and significant distance limitations.

Of course the most severe challenge to the future of DSL is the fact that in general it is a telephone company technology, empowered by the aggressive innovative environment common in telephone companies (NOT!). DSL is seen by them to be a way to provide mixed voice and data service over the same line so the version of DSL they are working on uses ATM to multiplex the services and is tightly tied to the voice world. In addition, the same people in the telephone companies who brought you ISDN are involved in bringing you DSL, a potentially fatal burden if there ever was one.

There seem to be dozens of types of DSL and not all of them have the same set of issues. IDSL and SDSL, the types of DSL that Copper Mountain and a number of other companies are using uses the same on the wire technology (2B1Q) as ISDN does which, among other things, means that the crosstalk problems that plague some types of DSL are not a significant issue. The ability to run at a reasonable speed (128 Kbps) over 22,000 ft of wire (faster for shorter cable runs) means far better coverage than DSL versions which have shorter limits. Over 99% of all customers are within 22,000 ft of a phone central office, this drops to less than 50% for 12,000 ft.

But the thing that seems most promising about the Copper Mountain approach is that they see this as a data service not some mixed media service. They are dealing with ISPs, not telephone companies, and use frame-based transport rather than ATM. Frame-based is cheaper and less complex to deal with than ATM.

DSL deployment is still very small, far less than cable modems for example, but developments like this and perhaps the "DSL-lite" under development by Microsoft and others, may mean that DSL will have a better future than it has a present.

disclaimer: Harvard has a long past of dealing with the future but has no opinion on DSL technology.