Overview

It seems like a lifetime ago, but I remember having an ongoing argument in 1989 about which was better: IBM's SNA protocols, DEC's DECnet and LAT protocols, or the up-and-coming TCP/IP protocols for commercial systems.

The discussion extended to many customers and many in the local DFWLUG user group joined the discussion with barbs and angry words sniping the other protocols. Async transmission is better then sync transmission it recovers better and let's you use lower-cost lines, peer-to-peer networking is the only way to go, standards are the way of the future, mine has more features, your protocol is lame, and so on.

In the end, all the arguments missed the point. We were all arguing the technical merits of peer-to-peer networking, counting the number of terminals that could be supported, or how many angels danced on the end of a 50-ohm cable. That wasn't the right focus at all, and many people missed the real reason for this change in networking technology in such a short period of time.

IBM's SNA could host tens of thousands of terminals in a single network with subsecond response time. DECnet provided peer-to-peer networking with over 100,000 nodes in a single network (when the Internet had almost 15,000 nodes), Novell was fastest for PC disk sharing, and the much-touted OSI protocol would unite all networks into a single, standards-based world that would correct all the shortcomings of past protocols. But all of the technological greatness of these and other protocols paled against TCP/IP.