In the beginning ARPA created the ARPAnet.
And the ARPAnet was without form and void.
And darkness was upon the deep.
And the spirit of ARPA moved upon the face of the network and ARPA said, ‘Let there be a protocol,’ and there was a protocol. And ARPA saw that it was good.
And ARPA said, ’Let there be more protocols,’ and it was so. And ARPA saw that it was good.
And ARPA said, ‘Let there be more networks,’ and it was so."
Danny Cohen - 1989
The American Department of Defence's Advanced Research Projects Agency Network (ARPANET) was the precursor to what we now know as "The Internet." Since its inception it has grown from a side project designed to keep lines of communication up in event of a nuclear attack on the United States, to something that revolutionized virtually all aspects of our lives. Especially yours. Yes, you, sitting there reading this. Ha!
Despite the fact that it was funded by the Department of Defence, the Advanced Research Projects Agency was geared towards pure research, with or without out any particular military application. It was founded partially as a reaction to the 1957 launch of Sputnik by the U.S.S.R.
At the time, the entire computer industry still thought of the computer as little more than a fancy calculator. I mean, who needs to use an extremely expensive computer for communication when you have the perfectly adequate, and cheap, telephone?
Most of the progress on computer networking had been done by ARPA's office of Information Processing Techniques. They had made some progress in the networking of computers that are the same, when the problem of sharing information amongst computers with different systems, etc came up. At ARPA's annual meeting in 1967, held at the University of Michigan, it was agreed that there were two main problems with a network of varied computers.
First they needed a 'subnetwork' of telephone circuits and switching nodes, with adequate capacity, reliability, and reasonable cost, which could help move along resource sharing among the network.
They also would need to design the protocols within the operating systems of the computers connected to the network, so that they could actually communicate with each other.
By March 1968, the specifications for the Interface Message Processor (IMP) were completed. Everyone now knew what hardware they needed to hook up the network. Someone just had to actually make it.
Bolt, Beranek and Newman (BNN) won the contract to build both the IMP's, and the network hooking them up together. There was still the minor detail of working out the protocol which the IMPs would use to communicate with each other, and how to hook up their own computers to interface with the IMPs.
It was decided to let the host sites work this out amongst themselves. This is likely because they knew a whole lot more about the type of computer they would be using than anyone else. Thus, they would be best suited to design the hardware and the software that would be required.
The sites for the first IMPs were chosen from places that had previously worked on ARPA projects, that could provide unique resources and network service support. UCLA, SRI, UCSB, and Utah were chosen to receive the first IMPs.
Leonard Kleinrock was in charge at UCLA, which received the first IMP on Sept 1st, 1969. It was he and his students Steve Crocker, Vint Cerf, and Jon Postel who first wrote the network protocol to make their Sigma-7 machine compatible with the IMP. Thanks to his work in queuing theory, he would actually be able to measure and understand what was going on with the network. As a result, UCLA was made the Network Management Center.
One month later, SRI received their IMP, and connected to the network. The first letters transmitted over ARPANET were L G O.
Meanwhile, Elmer Shapiro at SRI had been assigned the task of "making something happen," in regards to the actual protocol for communication. He called together a meeting, him, Steve Carr from Utah, Stephen Crocker from UCLA, Jeff Rulifson from SRI, and Ron Stoughton from UCSB. They threw together the DEL (Decode- Encode-Language) and NIL (Network Interchange Language). Soon after the first IMPs were installed, they had worked out this little thing called telnet. You may have heard of it.
IMP #3 arrived in Santa Barbra in November, and #4 arrived at the University of Utah in December.
At the time, the network was set up so that Utah was connected to SRI, SRI was connected to all three others, and UCLA and UCSB were connected to each other and SRI. This meant that if someone at UCLA or UCSB wanted to log into Utah, they had to go through SRI. If SRI was down, however, that was a problem. So, they needed to make a few adjustments to the subnetwork. While they were at it, they added more nodes. By Winter 1970, they had added MIT, RAND, System Development Corp., Harvard, and BBN. They had two high speed cross continent lines, one from BBN to RAND, and the other from MIT to Utah.
The next problem they worked on was the fact that they had no standard way of transferring files. So, some of the guys who cooked up this telnet thing, plus some researchers from some of the new nodes got together, and cooked up a file transfer protocol (ftp). You may have heard of it. It was done by July, 1972.
By this time, there were a lot more people on the network. I'm not going to bother listing them, but August 1972 saw the network with 29 nodes.
Spring of 1973, Vint Cerf and Bob Kahn began thinking about how to connect ARPANET with other networks. At the time there were two other networks they might have been able to connect to, SATNET and packet radio. These guys thought they needed a way to transmit data so it looked the same to each network. They worked out a protocol whereby data was sent in packets. It then checked to make sure that the packet was received, and if not, it sent a new one. The system was called Transmission-Control Protocol.
About the same time, Ray Tomlinson
worked out this little thing that became known as e-mail
. He decided to use the @ symbol to separate the usernames at individual nodes from the names of the computers at those nodes.
In 1977, Cerf became program manager of packet radio and SATNET. The three research programs were collectively called the ARPA Internet, because of the multiple internal networks. In 1978, Cerf, Postel and Danny Cohen at ISI decided to break TCP into two parts. TCP would take care of encoding and decoding the packets that it sent/received, and IP would be concerned with the transmission of these packets. You may have heard of TCP/IP. Or maybe not.
Throughout the late 70's and early 80's, there were more networks springing up, such as the Computer Science Research Network (CSNET), the CDnet (Canadian Network), and the National Science Foundation Network (NSFnet). Most of them were using TCP/IP, but ARPANET didn't start using it until New Year's Day, 1983. Some people view this as the birthday of the internet. These other computer networks continued to flourish, becoming to be used more for commercial purposes, and less for research. In case you haven't figured it out, it kept doing that for quite a while. Eventually, in 1989, ARPA decided that they couldn't keep up with funding to compete with the newer, faster networks. And so, Mark Pullen went to each and every node on the ARPANET, disconnecting it, and hooking it up to the NSFnet. Thus ended the 22 year existence of ARPANET.