15 ms·
Pretty perceptive for a 13 yr old. I was at Sun from the day they went public until the launch of Java, I saw things a bit more clearly. When Sun started they
by ChuckMcM 10y ago
Pretty perceptive for a 13 yr old.
I was at Sun from the day they went public until the launch of Java, I saw things a bit more clearly. When Sun started they were building workstations using Motorola's 68010 series of chips. They had taken an open source OS (4.xBSD) and a commodity microprocessor (68K) and put together some really nice hardware and made sure all the drivers worked and all of the software interfaces were consistent. Their two fiercest competitors at the time were DEC with its VAXStation line (custom CPU (VAX), custom OS (VMS)), and Apollo with its Domain line (commodity CPU (68k), custom OS (Domain)). It gave you almost the perfect Petri dish for evaluating the strategy.
It worked well, Sun out engineered all other workstation companies and by 1990 was at the top of its game.
The first thing that started eating at it was Silicon Graphics, the focus on serious graphics hardware got them into places where Sun was weak. The second thing was the slow ponderous rise of the x86 architecture. It wasn't until the 386 that there was realistically a competitor to the old 68K dominance in 32 bit machines. And the Pentium was the first architecture that could go head to head with SPARC on the low end.
At the top of its game Sun made three bad decisions, that ultimately sealed its fate. First it decided to make its OS more "enterprise" and got into a huge deal with AT&T to "merge" all the UNIXes into one happy System V. Second it decided that it could build a better CPU than semiconductor companies and embarked on building SPARC. And finally, when x86 machines started developing the same capabilities as workstations (and brought Windows NT along with them), it chose to retreat into the "datacenter" rather than continue to work and learn in deskop space.
All of the advantages of its early days had been reversed, it had a proprietary OS (Solaris), a proprietary CPU (SPARC), and it charged premium margins (even though they were not sustainable). I emailed Scott McNealy and suggested he sell off SPARC to Fujitsu, sunset Solaris in favor of a new version of SunOS based on the FreeBSD kernel, and engineer machines out of commodity chips (x86 and Opteron). They did much of that (except the SunOS part) but it was much too late and the rest of the organization was hemorrhaging money.
Even today I think that could be a credible business plan although I note that folks like System76 haven't been nearly as successful as the early Sun. I'm not sure if that is that they don't engineer enough of the systems or if they don't have the staff to really do the drivers they need. Or it could just be that crappy driver support and flakey window systems are the new normal and not enough people remember not rebooting for 6 months.