[ih] Protocol layering

Jack Haverty jack at 3kitty.org
Sun Dec 4 18:03:58 PST 2005

My recollection from the late 70s and early 80s is that we spent a lot
of meeting time trying to figure out how our TCP/IP/etc technology fit
into the OSI 7-layer model, with notions like "layer 2a" and such
shoehorning.   Eventually we gave up and just wrote code.

In retrospect, the CCITT model was based on telephony, and the only
major function that the telephone network really did was to create a
virtual circuit for streams of voice.   The Internet was so much more
complex that I don't think it fit well into that model.  Telephone
instruments only talked to each other and to "the system" when dialing
or disconnecting.   Internet "hosts" talk to lots of other hosts for
many purposes - web, file transfer, DNS, SNMP, ICMP, EGP/GGP/IGP, and
lately all sorts of other "P2P" stuff.   It seems to me more of a mesh
than a tree-like layer cake.

Hmmm, perhaps TCP/IP is the first example of the success of "Open
Source" - OSI docs were expensive and hard to get, TCP/IP was readily
FTP-accessible.   I suspect that most of the people writing the early
Internet code didn't have copies of the OSI docs.  (I had only a few of
them)  The prevalent attitude that I saw, both within the Internet
community and in the Telephony/X.25 crowd, was that TCP/IP was a
research experiment, and eventually the professionals would get the OSI
spec completed, implemented, and TCP/IP would be forgotten.

Then the customers decided to use what worked instead of waiting.

Anyway, if I was looking for an OSI document I'd try to find Mike
Padlipsky as a likely source of ancient documentation.

/Jack Haverty, Point Arena, CA
(ex-BBN, ex-Oracle)

On Sun, 2005-12-04 at 17:28 -0800, Dave Crocker wrote:
> > There are references to 1978 versions of both X.200 and 7498.  But I’ve 
> > not been able to find an actual copy of either. (Also, assuming a copy 
> > is available, do we still need to satisfy ITU/ISO distribution 
> > restrictions for these old drafts?)
> I seem to recall having specification of the original 7-layer model come out 
> in 1976, but perhaps that was not the final version of that round of effort.
> And as long as you have raised the subject, I will share with folks an 
> historical perspective that I've enjoyed about the effect that producing the 
> 7-layer model had:
> Before it existed, networking technical discussions among people meeting for 
> the first time usually took the first half of the meeting to agree on terms 
> of reference and the second half doing real work.
> The OSI 7-layer model was revolutionary.
> After it came out, similar first-time meetings consumed the first half, 
> complaining about the OSI model, and the second half using it.
> d/

More information about the Internet-history mailing list