7.1 Breaking the Bond

Free For All. Go to the Table of Contents. Vist the Gifcom.


The first move to separate Berkeley's version of UNIX from AT&T's control wasn't really a revolution. No one was starting a civil war by firing shots at Fort Sumter or starting a revolution by dropping tea in the harbor. In fact, it started long before the lawsuit and Linux. In 1989, some people wanted to start hooking their PCs and other devices up to the Internet, and they didn't want to use UNIX.


Berkeley had written some of the software known as TCP/IP that defined how computers on the Internet would communicate and share packets. They wrote the software for UNIX because that was one of the favorite OSs around the labs. Other companies got a copy of the code by buying a source license for UNIX from AT&T. The TCP/IP code was just part of the mix. Some say that the cost of the license reached $250,000 or more and required that the customer pay a per-unit fee for every product that was shipped. Those prices didn't deter the big companies like IBM or DEC. They thought of UNIX as an OS for the hefty workstations and minicomputers sold to businesses and scientists. Those guys had the budget to pay for big hardware, so it was possible to slip the cost of the UNIX OS in with the package.


But the PC world was different. It was filled with guys in garages who wanted to build simple boards that would let a PC communicate on the Internet. These guys were efficient and knew how to scrounge up cheap parts from all over the world. Some of them had gone to Berkeley and learned to program on the VAXes and Sun workstations running Berkeley's version of UNIX. A few of them had even helped write or debug the code. They didn't see why they had to buy such a big license for something that non-AT&T folks had written with the generous help of large government grants. Some even worked for corporations that gave money to support Berkeley's projects. Why couldn't they get at the code they helped pay to develop?


Kirk McKusick, one of the members of the Computer Systems Research Group at the time, remembers, "People came to us and said, 'Look, you wrote TCP/IP. Surely you shouldn't require an AT&T license for that?' These seemed like reasonable requests. We decided to start with something that was clearly not part of the UNIX we got from AT&T. It seemed very clear that we could pull out the TCP/IP stack and distribute that without running afoul of AT&T's license."


So the Berkeley Computer Systems Research Group (CSRG) created what they called Network Release 1 and put it on the market for $1,000 in June 1989. That wasn't really the price because the release came with one of the first versions of what would come to be known as the BSD-style license. Once you paid the $1,000, you could do whatever you wanted with the code, including just putting it up on the Net and giving it away.


"We thought that two or three groups would pay the money and then put the code on the Internet, but in fact, hundreds of sites actually paid the one thousand dollars for it," says McKusick and adds, "mostly so they could get a piece of paper from the university saying, 'You can do what you want with this.'"
This move worked out well for Berkeley and also for UNIX. The Berkeley TCP/IP stack became the best-known version of the code, and it acted like a reference version for the rest of the Net. If it had a glitch, everyone else had to work around the glitch because it was so prevalent. Even today, companies like Sun like to brag that their TCP/IP forms the backbone of the Net, and this is one of the reasons to buy a Sun instead of an NT workstation. Of course, the code in Sun's OS has a rich, Berkeley-based heritage, and it may still contain some of the original BSD code for controlling the net.