Alan J Sondheim on Fri, 18 Jul 1997 09:03:12 +0200 (MET DST)


[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

<nettime> Important Documents from the Early Internet (1972)


-


EARLY IMPORTANT NET DOCUMENTS (1972)



SCENARIOS for Using the ARPANET at the INTERNATIONAL CONFERENCE ON COMPUT-
ER COMMUNICATION (ICCC) from 1972 is the first guide to what later became
the Internet. The front page of this is illustrated in Peter Salus' Casting
the Net. The conference ran from October 24-26 in Washington, D.C. The 62-
page SCENARIOS was published by the ARPA Network Information Center of the
Stanford Research Institute, in Menlo Park, California.

A copy of the guide is with the BBN Library, and I have been tracking some
of the material down. The ICCC was the first real introduction of the Net
to the (professional) public, and this then becomes the first how-to, with
examples indicating the thinking of the Net community vis-a-vis usage and
implementation. This is the sociology, not the technical history of the
Net, then. At the ICCC, there were a number of terminals; participants
could walk among them, and log in. Each "scenario" connected to a differ-
ent application, located somewhere across the United States.

The following, which I have now in xerox form, are the programs and the
scenarios:

PROGRAM                         SCENARIO

English Language Conversational Programs
DOCTOR                          BBN DOCTOR     [Eliza-type program]
SCHOLAR                         SCHOLAR
PARRY                           SAIL PARRY
TIMMY                           UCLA-NMC Sigma-7

Data Base Query
NIC                             SRI-ARC  ["general intellectual tasks"]
NETWRK                          MIT-DMCG PDP-10
APE                             SAIL AP HOTline    [Associated Press]

Games
CHESS                           BBN CHESS
CHESS                           MIT-AI PDP-10
LIFE                            BBN LIFE     [Conway's game]
JOTTO                           MIT-AI PDP-10

Network File Transfer                   [developed later into ftp]
SMFS                            SRI-ARC
RJS                             Remote Job Service

Miscellaneous
ABACUS                          UCLA-NMC Sigma-7
HELP                            UCLA-NMC Sigma-7

Programming Languages
SPEAKEZ                         SPEAKEASY
PPL                             HARVARD PDP-10
FORTRAN                         BBN Tenex
FORTRAN                         UCLA-CCN 360/91 TSO

Remote Job Entry
RJS                             Remote Job Service

Symbolic Algebraic Manipulation
MACSYMA                         Mathlab's MACSYMA

There is a table of contents, which also lists MIT H645 Multics, which
apparently provided mail; mail was also available through BBN Tenex
itself. For those not familiar, BBN is Bolt Beranek and Newman, Inc.,
responsible for the first IMP, basically routers, that were the founda-
tion of the Arpanet. IMP = Interface Message Processor.

SRI-ARC (NIC) was set up at HOST #2 in the room. Control characters were
indicated by an up-arrow sign, which later must have transformed into the
current ^. For example, Control-a allowed backspace with delete; Control-
c returned control to the TENEX EXECUTIVE SYSTEM (which roughly parallels
Control-c for ending execution in Unix, linux, or DOS); and Control-t
"checked to see if the system is still there," similar to the "are you
there" command in some telnet programs.

Other commands included, of course, DEL, CR (carriage return), etc. Com-
mands were preceded by @.

The SAIL AP Hotline was at HOST #11, "a direct Associated Press news line
carrying national and international news. The AP Hotline has been
interfaced to the SAIL system at the Stanford Artificial Intelligence
Laboratory. Any terminal on the ARPA Network can be turned into an AP news
line by running program 'HOT' at Sail."

The following setup had to be performed first:
echo remote;
insert linefeed after every carriage return
login to SAIL
LOGGER says you're being connected
T R OPEN says you're connected, both transmit and receive.

Then a second login opens the particular program; the word is "hot" with
some other commands:
r hot <CR>
(in other words, type r, space, hot, carriage return).
You then get the linefeed from AP until you type...
Control-c.

Finally, let's look at the UCLA-NMC HELP at HOST #1. The description is
probably that of the first on-line example of hypertext (and very very
early hypertext as well):

"HELP is a subsystem at UCLA-NMC which permits a user to interrogate a
database which is organized in directed graph form. Each vertex of the
graph has a paragraph of information, including some information about
further details which can be obtained from vertices which are reachable
from the current one.

"Thus, the user moves from vertex to vertex, investigating each item as his
interest directs."

The login sequence is six steps. You then login:
LOGIN iccc <CR>
and get a return:
JOB STARTED
Then you type
help <CR>
and get
NNN HELP STARTED
where NNN ps the PID, or process number assigned to HELP.
You next read
DO YOU KNOW HOW TO USE THIS PROGRAM?
no <CR>
and you get a tutorial.
Or you can answer yes, and then read:
ENTER A SERVICE NAME, X, or ?
to get a list of things you can get help about...

Here are the help functions:
HELP - short description
LOGIN - ditto
MSG - "how to use our message processor"
NETWORK - "tutorials on network resources"
SRVYGRAPH
TELNET        -       note this early appearance of TELNET for the public!
SURVEY - "some random comments"

-----------------------------------------------

One of the critical papers presented at this ICCC is "THE NETWORK CONTROL
CENTER FOR THE ARPA NETWORK," which monitored the IMPs. It was written by
Alexander A. McKenzie, Bernard P. Cosell, John M. McQuillan, and Martin J.
Thorpe. I have a copy of the preprint.

Here is a list of node from it, from TABLE 2: SUMMARY OF NETWORK OPERATION

May 1971        15 nodes
June            15
July            15
August          15
September       18          2,892 Average Host Intersite Output
October         18          5,329 (packets/node/day)
November        18          6,473
December        19          5,679
January 72      19          9,055

The average line outage during these segments ranged from .59% to 3.21%.
The average IMP downtime ranged from 1.77% to 5.50%.

Here are some of the reasons for the IMP downtimes:
Main frame problems, repair, unknown.
High voltage, test, repair power transients.
Intermittent core stack IMP move.
Software, repair, site Host test.
Site power down.
Blown fuses.
Software bug.
New system reload, NCC error. [NCC = Network Control Center]

Then as now, the questions at the end of the paper concerned issues of
bandwidth:

"What are the peak hours of network use and what is the peak-to-average
traffic ratio?
"What percentage of network traffic do single-packet messages constitute,
and how does this percentage vary from Host to Host?
"What is the ratio of weekday use to weekend use?
"What percentage of line capacity is used during peak hours, and during
weekend?"

Note the oddly religious overtones from the capitalization of "host."
Processing of information was done by paper tapes; BBN had another Host
for dealing with the tapes, which was going to be used for analysis as
well.

Finally, here is the abstract to the paper:

"The ARPA Network allows dissimilar, geographically separated computers
(Hosts) to communicate with each other by connecting each Host into the
network through an Interface Message Processor (IMP); the IMPs themselves
form a subnetwork that can be thought of as a distributed computation
system. To detect failures in this system each IMP automatically and per-
iodically examines itself and its environment and reports the results to
the Network Control Center (NCC), at Bolt Beranek and Newman, Inc., for
action. The NCC computer, like any other Host, can itself fail without
affecting network integrity; further, the NCC central processor can easily
be replaced, in case of failure, by any standard IMP.

"The present paper briefly describes the NCC hardware; discusses such
software issues as NCC-related routines in the IMPs, data-collection and
interpretation mechanisms, line status determination, IMP status and pro-
gram reloading, and Host and line throughput; details NCC operations (man-
ning, problem-handling procedures, track record); and summarizes overall
NCC experiences and future plans."

Thanks to Janet Abbate of the Center for the History of Electrical Engin-
eering; Alexander A. McKenzie; and BBN itself.

Note in relation to a timeline: the first four nodes connected in late
1969; the growth was slow at first, as the machines were tested. These
nodes were UCLA, UCSB, SRI, and UTAH.

_________________________________________________________________________

---
#  distributed via nettime-l : no commercial use without permission
#  <nettime> is a closed moderated mailinglist for net criticism,
#  collaborative text filtering and cultural politics of the nets
#  more info: [email protected] and "info nettime" in the msg body
#  URL: http://www.desk.nl/~nettime/  contact: [email protected]