Name: David Bernstein Age: 36 Place of Birth: Orange, CA Position: Vice President of Engineering, Reach Software Years in Current Position: 1 Years in the Industry: 13 Car He Drives: 1990 Thunderbird Super Coupe ("It's red, and I always speed.") Last Book Read: "The truth is, I can't remember the last time I seriously read something that wasn't on-line, except to my kids." Pet Open Systems Peeve: Marketing people use the word open but the end user, the distributor, the reseller, or the systems integrator can tell you what open means and that is good enough. Open is not a marketing name; it means that it provides you with the price/performance, the benefits from choice, and the flexibility that you need. If it does, then for you it's open, whether it's called open or not."
For the young David Bernstein, building a computer came as naturally as building a tree house did for other kids. The son of an engineer and an educational product of the Silicon Valley of the 70s, Bernstein followed his scientific bent into a career in computer software. In recent years, he was instrumental in The Santa Cruz Operation's participation in UNIX unification efforts such as The Common Open Software Environment (COSE), Common Desktop Environment (CDE) and UNIX International. Today he's vice president of engineering for Reach Software in Sunnyvale, CA, a provider of workflow software.
Bernstein's father, a metallurgical engineer, moved from Southern California to the San Francisco Bay Area when David was young to work for some of the pioneering semiconductor firms, including Fairchild, National, Signetics, and Hewlett-Packard. "I got exposed to electronics," Bernstein recalls. "It was just part of what you did. We always had lots of TVs and stereos and I used to take apart telephones and radios. I built a kit computer from Popular Electronics when I was in high school. The school [in Cupertino] had tons of surplus equipment from all the companies in the valley. Most of the kids in my neighborhood had dads who worked at some semiconductor company. I was a Silicon Valley kid and it didn't seem like there was anything else to do."
Part way through high school, Bernstein's family moved to Santa Cruz, a beach town 30 miles from Silicon Valley, where a new campus of the University of California had sprung up. One day in 1974, he and some friends decided to go to the university campus, nestled in the redwoods, and have a look around. "They had just installed some DEC PDP11s running a very early version of UNIX," Bernstein remembers. That was really one of my first software experiences because it was a real computer. Providentially, I guess, it was my first exposure to UNIX." Bernstein would later graduate from UC Santa Cruz and become a developer and proponent of UNIX.
Bernstein put that physical science career on a sidetrack when he took his first job after graduating in 1981. "I knew UNIX and got a consulting job with a little local group of half a dozen guys who just did UNIX consulting," Bernstein says. That group of guys was the nucleus of The Santa Cruz Operation, the company that today is the leading provider of UNIX operating systems for Intel-based PCs. "They were just starting out that year," Bernstein says. "We were just consultants for hire. We would write device drivers or do consulting on UNIX systems or anything that would make a buck. It seemed to me like something good to do before I got a real job. And it just took off like crazy."
The company's expertise in UNIX grew, as did interest in running UNIX on early PC hardware. A number of companies, including Zilog and Onyx, produced PCs based on Motorola's 68000 microprocessor, running some version of UNIX. At that time, AT&T would not license the name UNIX but would supply the source code, so companies came out with their own versions. When IBM came out with the Intel-based PC running DOS, other companies tried to clone it. "I remember Victor Computers, which was in Scotts Valley [near Santa Cruz] trying to make a PC clone," Bernstein says. "They were going to do a better job than IBM by supporting not only DOS but UNIX. They gave us the contract to do the UNIX work.
"Eventually, Victor went under, but we thought UNIX on Intel PCs was a pretty neat idea, and we thought other people would. It certainly was better than DOS. It was on a whole different level. You could do multiuser and multitasking. Because Victor's machines weren't ready, we had used the IBM PC as a reference machine. It really was a very bad UNIX machine, but a few other guys and I got UNIX to work on it and we realized we had done something pretty neat. [In about 1984] SCO made a big announcement, when we had the three leading personal computers of the day-the IBM PC, the Apple Lisa and the DEC Pro 350-all running a version of UNIX from us. We were all very impressed with ourselves. We got the idea that we were now going to be the UNIX provider for PCs."
SCO's hopes were further boosted by IBM's earliest version of the PC AT, based on Intel's 286 chip, which was to run an advanced operating system. "The 286 was designed by Intel to run a protected-mode operating system, and it wasn't going to be DOS," Bernstein says. "When you bought a PC AT you got UNIX on it from Microsoft. It was called Microsoft XENIX. Microsoft was at that time the leader in the UNIX business. We thought that, now that IBM was selling the PC AT with UNIX, we were there, we were dialed. But it turned out there were hardware bugs in the PC AT, so people ran DOS on it for awhile. Microsoft lost its chance with UNIX and UNIX lost its chance on the PC."
SCO, however, still had plenty of business. "In the meantime, companies were coming to us for consulting. Microsoft got IBM's business and gave the rest to us, so we just kept cranking out those UNIX ports and turned our product into a retail product." SCO called its first UNIX product DYNIX, but sold that trademark to Sequent. ("We thought it was great that we could get money and not do any work.") Needing another name, SCO licensed the XENIX trademark from Microsoft for "a great price. That's how SCO got hooked up with Microsoft in this strange story."
SCO eventually found its niche through its early device driver work. "When we were in the earlier business doing work on PDP11s and VAXs, there was quite an aftermarket in PDP11 and VAX add-in cards. You could buy a disk controller from Emulex instead of Digital and it would be faster and cheaper. A lot of people asked us to write UNIX drivers for third-party peripherals for DEC systems. When we saw the PC architecture come out, we said, 'Wow, we're going to get a lot of consulting deals for writing drivers for third-party cards.' In fact, we thought this whole add-in market was going to be very important. So we started calling up hardware people and we did a whole bunch of device drivers for their ports. A lot of the VARs and system integrators came to us saying they wanted to mix and match boards and machines, and could we give them a package of UNIX that has all the drivers. We turned that into a product which really sold. That was SCO UNIX, first for the 286, and then for the 386."
From the start, SCO's idea was to enable customers to build systems by mixing and matching parts. "To us, the concept of open systems started with saying that customers can build their own computers out of standard components, and now they don't have to buy the software from the place they bought the hardware." The application side of the company contributed to the concept by porting various popular applications for customers, allowing them to buy those off the shelf to complete their systems. It was "absolutely ridiculously hard" to convince users that they actually could build their own systems from parts, Bernstein admits. "We spent years on the brink of extinction because of that. Luckily we weren't alone. UniForum and other organizations were very important to us because we weren't the only ones trying to give this pitch.
"The two most compelling arguments were first, the cost/benefit equation-you could do the same thing for less if you could load it yourself. Second was freedom of choice. For example, Digital had some absurd policies. If you put somebody else's board into their system it would invalidate your warranty. People thought that was onerous. The reason people started buying these small computers was to get away from the absurd policies of the minicomputer and mainframe computer people. Fundamentally, the market wants choices."
Today, Bernstein is still active in communicating the open systems message as a member of the 1995 UniForum Conference Planning Committee.
It was also to further open systems that SCO sent Bernstein to work on initiatives such as the Advanced Computing Environment (ACE) and COSE/CDE, and in multivendor consortia such as UNIX International and the Open Software Foundation. "Remember, all this time UNIX was a fundamentally superior operating system," he says. "Its whole multiuser thing and its networking were way ahead. But this is not to say that SCO succeeded as much as we should have. In the end, we made a tiny, tiny dent in Microsoft's rise to power. So it's not like we really succeeded. But basically, the core technical teams stuck to the original goal, which was to get the most out of the machine. That was not the DOS way-making the fewest changes you can get away with."
The second event he likes to remember happened just last year, after UniForum '93, when Sun CEO Scott McNealy agreed, as part of the COSE initiative, that Sun would embrace the Motif graphical user interface and supply it for Sun developers. "When the dust settled behind the scenes of COSE, Sun found itself in a window between UniForum '93 and their own developers conference, when they suddenly needed to deliver Motif to their developers. They had never looked at Motif, and I knew there was no way they could build one for Sparc that was going to be quality. SCO owned IXI, and I licensed Sun the IXI version of Motif. It was one of those lifetime OS achievements. That was sweet revenge, I guess."
Bernstein moved to Reach when "I actually thought that the operating system was becoming less important than a lot of the networking and services infrastructre coming out. I thought Lotus Notes was very important and I thought electronic mail was very important. So I joined a company whose products run on Windows and that uses electronic mail and Lotus Notes. It was a chance to join a startup again, with under 50 people, and run an engineering team again, and try to build some products in these exciting new areas. What I've learned in the open systems business has been completely invaluable in working in these other areas.