While working with UNIX at Amdahl, I had been up to Berkeley a couple of times for UNIX seminars and I had met Bill Joy there. When Sun got started in 1982, Bill gave a list of known UNIX hackers to Scott McNealy to start trying to recruit. However, it was a secret until later that Bill would be leaving Berkeley and joining Sun. So one day at Amdahl I got a message slip saying that some guy named Scott McNealy had called.
It turns out I had heard through friends about the S.U.N. hardware that Andy Bechtolsheim had designed at Stanford, so when Scott mentioned that they were building a company around it I decided to interview and learn more. Now at that time, there were already about 100(!) separate startups in the valley doing 68000 processor based UNIX systems. The 68000 was one hot micro and UNIX was very well known to the graduating college crowd, so there was a lot of buzz going on. What set Sun apart was 2 things – a desire to build graphical workstations, like the Xerox PARC Alto, and a desire to advance the state of UNIX, which was what Bill Joy had been doing for DARPA at Berkeley. So by the end of 1982 there were a zillion little 68000 UNIX box companies, but they were all based on Version 7 UNIX with dumb terminals and had severe limitations like no virtual memory. A lot of the culture of the 68000 UNIX crowd was coming from the home-built personal computer guys, who could make cool stuff but weren’t known for product completeness and reliability. The Sun culture, on the other hand, came
from people who were used to “real” computers, Bill Joy and Bill Shannon with VAXen, and myself with mainframes.
I started at Sun on May 3, 1982. There were various people running around in chaotic ways, but a fellow named Bruce Smith seemed not to be too busy, so I followed him around asking questions. As it turned out, it was his first day too, and he was hanging around me because he thought I knew what was going on!
I think we shipped the first Sun-1 workstation on June 1st, so obviously there had been a lot going on before I got there. My major contribution was to fix the Interphase disk driver that came with Unisoft UNIX so it didn’t scribble randomly on the data. However, even the Sun-1 wasn’t Sun’s first revenue product! It turns out that Andy had agreed to build some Multibus cards for 3 Megabit Ethernet for Xerox. Xerox had plenty of their own Ethernet cards, but not for Multibus, so Andy as Sun sold about 30 to them.
Like lots of other companies, we used a 68000 UNIX port from
Unisoft to get to market, but we didn’t plan to stay with it. The VAX world at
that time was running 4.1BSD, but the 4.2BSD project specs had been widely
circulated and everyone knew it was coming “real soon”. This crowd was shocked when
Bill announced that he was leaving Berkeley to join Sun, but he promised to
help finish 4.2. Sun’s first BSD UNIX, SunOS 1.0, was based on BSD 4.1c which
itself had pretty much everything that the ultimate 4.2 had. 4.1c had the new
virtual memory system which we married with the new 68010 microprocessor which
was the first to support real page faults. John Seamons at Lucasfilm was the
beta customer – he had been messing with SUN hardware before Sun came along.