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.

A lighthouse as a metaphor.

By Scott Bradner

Jules Verne's last novel, "The Lighthouse at the End of the World", is about a lighthouse built in 1884 on the island of Isla De Los Estados just east of Tierra Del Fuego at the southern tip of South America. The lighthouse was designed to warn the ships trying to make it from the Atlantic to the Pacific about the dangers posed by the island. The lighthouse was operated by the Argentine military until the rough conditions on the island forced its abandonment in 1902.

Earlier this year a French navigator, who had read the book as a child, completed a reconstruction of the original lighthouse in its original location. It had taken the navigator more than 4 years to raise the funds and, working with a crew of 7, more than two months to finish the construction. On February 26, 1998 the lighthouse was donated to the Argentine Navy which will operate it even though it serves no current practical value since satellite navigation systems are used by just about all ocean going vessels these days. Also, since the island is now a sanctuary where the public is not permitted to land, the lighthouse is only viewable from the occasional passing ship. The New York Times report on the occasion noted that an Argentine Navy Captain talked about the "symbolic value of the lighthouse." in that it represented the "dreams of explorers."

What brought this story to mind was the article on ATM flow control (ATM flow control: Recipe for demanding apps) in the August 31st issue of Network World. The article described ATM available bit rate (ABR) flow control and included a schematic of an ATM network complete with ATM to the desktop. But I'm not sure if people should be interested in this topic.

Don't get me wrong. I think that ATM ABR is quite a technical achievement and, if ATM were a common end to end networking solution, could be quite important. But very few observers think that ATM will play a significant role in connecting desktop computers to the rest of the world. ATM is just too expensive and complex for almost all locations. The dim prospect for ATM to the desktop is an important factor here since ATM flow control was designed to function end-to-end. It is not at all clear how to use ATM flow control with Ethernet or token ring connected machines..

Clearly ATM is now, and will likely to continue to be, an important WAN service. It could become quite important indeed if the regional phone companies are actually able to deploy inter-LATA, distance insensitive, ATM services. ABR could be of use in these cases if reasonable ways can be developed to transfer the flow control from the WAN to the LAN segments of the connection.

So at this point I am not sure if all of the dreams of the technology explorers that has gone into ABR have produced a symbol of technical achievement with no practical value or if ABR will help support useful ATM-based services.

disclaimer: Harvard has been claiming to have practical value for a rather long time but does not have an opinion on ABR.