The following text is copyright 1997 by Network World, permission is hearby given for reproduction, as long as attribution is given and this notice is included.
Who is wearing the blinkers?
Reality seems to be quite different depending on where you stand.
Clearly the general assumptions about the future role of ATM in data networking have changed significantly over the last few years. Only a few years ago the majority of pundits seemed to feel that ATM was going to take over the data networking world. It would become ubiquitous and replace all current technology. There were a few of us weirdoes that disagreed and were regarded with a touch of scorn but so it goes in the pundit biz.
Various things conspired to make ATM's path to ubiquity more than a little rocky. The price of 10 Mb Ethernet switches dropped precipitously, 100 Mb Ethernet became widely available, Gigabit Ethernet started peeking over the near horizon, the ATM standards efforts lagged well behind expected time frames and the early ATM systems turned out to be quite a bit more complex than the network people had expected. These factors have all but eliminated any idea that ATM will take over the world but there are still two rather different views about ATM's future and I'm having a hard time figuring out which one is closer to reality.
I have felt all along that ATM was going to be one of a number of important networking technologies in the data networks of the future. I have felt that very high speed wide area links and corporate backbone networks were most likely to be implemented with ATM. Also, for the foreseeable future most large corporate intranets were going to be a mixture of ATM and other technologies with most LANs being Ethernet and low speed WAN connections using frame relay.
But I am now beginning to hear another view, and not just from the radical IP community. More and more of the technical people I talk to are starting to relegate ATM to the dustbin of history as yet another technology that missed its window of opportunity. Their claim is that alternative technologies are in place or imminent for whatever features of ATM that network designers actually need. 100 Mb Ethernet has already just about eliminated ATM as a viable desktop technology and Gigabit Ethernet is doing the same for the campus backbone. In the wide area the recent announcements of very high speed frame relay capabilities with 155Mb available soon and 622Mb within a year would permit the telephone companies to provide connectivity services with link speeds that only ATM and SONET were able to previously support. (For a number of financial and physiological reasons many phone companies have been reluctant to enthusiastically market SONET.) They also claim that the type of quality of service that ATM provides is not what is needed in real world networking, particularly where it will not be end-to-end (Ethernet at the ends) and where you can throw bandwidth at the local campus backbone.
I still think that ATM will be a part (size to be seen) of our networking future but am starting to worry that maybe I'm just wearing blinkers and do not see the handwriting on the wall. Or is it the ATM-is-dead crowd?
disclaimer: Harvard asserts that it understands reality so the above confusion must be mine.