Kathryn Britton is an SNA architect from IBM Raleigh (I think) who undertook to explain away the fundamental differences between LU2 (3270) devices and LU6.2 (APPC). In a nutshell, here it is: under LU2, the host is smart and the terminal is stupid. In LU6.2, both ends are smart, and neither side "controls" the conversation.

While I can't complain about Kathryn Britton's presentation, or her depth of knowledge, I have a minor problem with these architecture types that IBM sends out to preach the gospel. These people spend a lot of time massaging the gray matter in front of blackboards, but not much time slugging it out in the salt mines with the rest of us. It is easy for them to say the future is bright and rosy with LU6.2 -- all you have to do is (wave of hand) upgrade your systems. These people don't have any concept of the inertia that a large 3270 network has.

LU6.2 is a half duplex protocol, and this has been the major complaint from the unwashed masses. Kathryn Britton says "We do understand the requirement for full duplex".

There is no such thing as RTM for LU6.2 connections!