supercomputers. A full spectrum of cutting-edge
computer science research. So the ARPANET
project is not viewed as a communications
thing, but a natural evolution of extending the
interactive computing environment as a way
of knitting together all of these laboratories so
that they can exchange information, so that the
computers can cooperate on the same problem
if necessary, so people can collaborate.”OPEN GATEWAY
BBN’s plan made use of the IMPs as gateways that
connected local resources such as mainframes.
They handled all the network calculations and
were connected using leased lines via modems
at 56kbit/s. The first-generation IMPs were
ruggedized Honeywell DDP-516 computers,
each with 24KB of magnetic-core memory. The
16-bit DDP-516 operated mostly at 2.5MHz, with
some parts able to reach 5MHz, and was the first
mini-computer to retail for less than $10,000.
In its rack-mounted configuration, it weighed
120lbs and drew 475 watts, but the IMP was housed in a metal
cabinet the size of a refrigerator with 24 indicator lamps on
the front showing the state of the network channels. Despite
being introduced in 1969, a DDP-316 was still operating in 2000,
monitoring the reactor core temperature at a nuclear power
station in Essex, United Kingdom. It only stopped working when
its internal 160KB disk failed.
Each IMP could communicate with four local hosts and six
remote IMPs via leased telephone lines, and the first four nodes in
1971 were UCLA, Stanford Research Institute, UC Santa Barbara,
and the University of Utah School of Computing. The abortive first
message, ‘LO’, was sent from the SDS Sigma 7 computer at UCLA
to the SDS 940 at Stanford. A student programmer, Charley Kline,
attempted the link at 10.30pm on October 29th, 1969 but the firstsuccessful login was subsequently achieved
about an hour later.
A permanent link between the IMPs at UCLA
and Stanford was established on November
21st, 1969 and the full four-node network was
complete by 5th December the same year. What
it didn’t have yet, however, was much in the way
of network applications, and used something
called the 1822 Protocol, named for the BBN
report number that specified it. Messages sent
by 1822 used numeric network addresses and
were guaranteed to be transmitted reliably. In
the event of a failure, a message was sent back
to the originating machine indicating there was
a problem.
The Network Control Program came next,
providing the middle of the network stack and
a firmer base for things like file transfer and
email. The @ symbol at the center of email
addresses was chosen by programmer Ray
Tomlinson in 1972.
Crocker led NCP’s development, while other
members of the Network Working Group worked on TELNET,
FTP, and SMTP. In March 1970, an IMP at BBN in Cambridge,
Massachusetts, was connected to the network, and by September
1973 there were 40 nodes. A satellite link to the NPL network
in London was considered, but University College London
was chosen instead. In June 1973, Kahn’s SATNET connected
ARPANET to the Norwegian Seismic Array (NORSAR, used to
detect nuclear explosions through seismic waves), via a downlink
in Sweden, then via cable to UCL and, eventually, to NPL, the first
time two distinct networks had been connected. Internetworking
was go, albeit at 2.4kbits/s.
In an interview in 1988, JCR Licklider, head of the Information
Processing Techniques Office at ARPA, remembered the
difficulties of dealing with the British on the project: “When I wasThe Internet Hall of Fame Inductees for 2012. Vint Cerf is fourth
from right in the back row, Steve Crocker is fifth from right
in the front row, with Leonard Kleinrock two seats to his left.
Robert Kahn is represented by Patrice Lyons, next to Cerf.Steve Crocker speaking in 2012,
while he was chairman of ICANN.APR 2022 MAXIMU MPC 43
© WIKIMEDIA, GETTY IMAGE/MIGUEL ROJO / STRINGER