Feed on

On my Symposium List this morning - a discussion thread building steam on whether list members should spend time keeping up to date on IMS which is generally greeted with scorn. Fred Goldstein commented on ATM and Andrew Odlyzko replied:


Thanks a lot for the history piece.

But wasn’t the basic mistake of ATM that it was supposed to provide desktop-to-desktop connectivity using switched (vitrual) circuits, and that turned out not to be needed?

When I first got into economics of data networks in 1997 (prior to that my involvement with networks was limited, and primarily on mathematical algorithms), I went around AT&T and talked to the folks in various units, such as those devoted to ATM, FR, private line, and Internet. (I seemed to be the only one doing that, strangely enough.) And the biggest frustration that the ATM and FR folks had was that their customers did not care about all the fancy features that those networks had. What those customers wanted was basically to transmit IP traffic from one site to another (or to the public Internet cloud). Basically IP had won the desktop and local connectivity competition, as it was good enough and flexible enough. (As just one example, when the NSF Internet backbone was getting phased out towards the end of 1994, the AT&T internal network that was carrying billing records for the voice network apparently was about as big as the NSF backbone, and had been transitioned to IP.)

Trackback URI | Comments RSS

Leave a Reply