[ih] Hourglass model question

[Joe Touch]

Hi, all,

On 7/3/2019 10:20 AM, Jack Haverty wrote:
> My recollection from many meetings in the early 80s of the ICCB/IAB, TCP
> Working Group, Internet Working Group, et al is that the TCP/IP
> community did *not* borrow the OSI model at all.  Or at least not in the
> early days of the Internet through 1983 or so, I'm not sure what
> happened later.

I think there was at least some terminology borrowing; not sure who came
up with what first, e.g., link, net, transport, etc.

However, the idea that specific functions were unique to a single layer
was (AFAICT) born and died (thankfully) with OSI.

The idea that there are a fixed number of layers or that they are
absolute, rather than relative, died with tunnels, IMO.

The (correct) idea that a layer has defined interfaces above and below
was almost killed by the IETF (in declaring that APIs were
"implementation issues" for a long time), but has (thankfully) since
been recognized as important.

And of course, John (Day) and I both believe layers are really instances
of a single, recursive concept.

YMMV.

Joe