The following text is copyright 1995 by Network World, permission is hearby given for reproduction, as long as attribution is given and this notice is included.

Is ATM a Synonym?

By: Scott Bradner

Some people seem to think that I'm anti ATM.

I guess that comes from the fact that I have often expressed the view that ATM is *an* answer but not *the* answer to the future of data networking. Please understand that I do not say "data networking" to reduce the extent of the application of this type of technology. However, I feel that within the next 5 to 7 years the phone companies will have to transition from today's environment where data, to them, is a special case of voice to a new environment where voice will be a special case of data. Keep in mind that, when I say data I include everything from text files to video playback. I think that there will be reasons to use a variety of networking technologies in the data networks of the future. Personally, I think that the big players over the next five years will be 10 Base T and 100 Base T Ethernet, Frame Relay, ATM and SONET. I also expect that a new next-generation networking technology, which many in the trade press will label as ATM's successor, will start to be identified in about the same time frame.

This column may not help to dissuade those of you who dismiss me as some sort of heretic but here it goes anyway. I've started to notice an interesting trend. I do quite a bit of consulting and often get asked to review organization's networking plans. More and more of these plans and more of the presentations of these plans are talking about ATM without talking about ATM.

They talk about "planning for ATM," or "getting ready for ATM" or "keeping ATM in mind". But when you start talking details it turns out that many of them are not talking about ATM because of ATM's planned ability to control quality of service (QoS), or because they see a general requirement for real-time applications like desktop video conferencing. They are using the term ATM as a synonym for *fast*. Clearly there are reasonable expectations in some environments for real-time applications, but most of the people I'm talking to these days are looking for speed and, perhaps, the ability to create virtual LANs.

They see a need for fast networking technologies in the future and the one that is on the tip of everyone's tongue is ATM. However, this is not because of any characteristic of ATM other than the widely touted transfer speed and some, often vaguely understood, ability to create LANs that are not limited in physical configuration.

Sure ATM can go faster than OC3 but OC3 (and even slower) speed ATM is what is affordable, or at least can be predicted as becoming affordable in the near-term future and with OC3's 136 Mbps or so payload capacity is not much different than 100Mbps fast Ethernet.

There are a growing number of 100 Mbps Ethernet switches. Some of these already support some form of virtual LANs (with more coming) and full duplex data paths. These devices are providing network planners with what many of them need from ATM without having to leave their comfortable Ethernet environment.

I'm not sure exactly what this portends. It does seem to me that some chunk of ATM's customers may disappear by the time ATM is ready to support them.

disclaimer: The quality of Harvard's service is, of course, excellent, so any discussion involving it must reflect my own opinions.