9 Jan
2008
9 Jan
'08
4:11 a.m.
Richard wrote:
In article <4784208A.8589EB19 at cs.ubc.ca>,
Brent Hilpert <hilpert at cs.ubc.ca> writes:
Yes, ethernet was precisely what I had in mind as I added that little "at least
above the link level" clause to my comment.
Ethernet is quite distinct from TCP/IP, it's just another link type as far as
IP is concerned. DEC latched on to ethernet quite early to use as a (choice of)
link layer for DECnet.
It is kind of funny that manufacturers could agree on physical & frame level
stuff (RS232,HDLC,ethernet,etc.) but then had to go their own way in the higher
layers.
Or one might say that agreement started at the physical level, and it just took
time to go up the stack getting agreement along the way. (Or else it's just
easier to be proprietary in software than hardware.)
I well remember linking VAXes together and fighting with 2-metre loops of
thick-wire ethernet while banging those damned metal brick transceivers around,
until DEC saved us all with that first 8-port ethernet 'hub' (although the
transceiver cables were still a pain). (Am I correct in recalling it was called
the DEQNA?).
Youngsters out there should be grateful they don't have to wire their home
machines together with thick-wire...
There's a lot of room for shades of gray in
there (what does 'forced' mean, a
nd
how far, and when) but I'd say there is a
good deal of validity to that
argument. In the 80's DEC and IBM were pretty big on pushing their own
'networking solutions' and did their best to avoid open protocols, at least
above the link level. TCP/IP didn't have the degree of prominence in the
networking world in those days as it does today.
But wasn't DEC instrumental in helping ethernet succeed? See the
history section of <http://en.wikipedia.org/wiki/Ethernet>. Or am I
confusing ethernet (physical layer) with tcp/ip (protocol layer)?