Ken Olsen, Who Built DEC into a Power (2011)
nytimes.com
nytimes.com
Another thing to take away from DEC is motivation and process. DEC was a company made of engineers who built things engineers wanted to use. Such companies still wow crowds in headlines today and is a niche worth remembering. The other part was the quality-assurance regiment that I heard they used for VMS. They'd code new features for a week with tests, let tests run over weekend, and then spend entire, next week fixing problems rather than developing new features. This alternation made for some really robust code. They similarly payed close attention to design and operations with the collective result being systems that lasted years under load without reboot.
So, management failures & desktops aside, Ken Olsen and his company DEC exemplified some good, lasting lessons for creating and marketing high-quality systems. Worth remembering both for this.
Just to provide some more contrast... Just about every other system had to be rebooted daily or every few days.
The problem with VMS was the hardware. Eventually the pain of being on slow, expensive machines allowed other companies to take more share. By the time the Alpha systems were ready, the Pentium (and soon Pentium Pro) was suprising every OS vendor.
That's the biggest point I push on it: architecture and development processes that worked in terms of reliability. It didn't emerge after billions of investment in whatever wasn't working. Just teams working 9a-5p building stuff, testing it, reviewing it, polishing it. And it totally worked.
"By the time the Alpha systems were ready, the Pentium (and soon Pentium Pro) was suprising every OS vendor."
People of the time period told me it was a combo of hardware and licensing cost. They said OpenVMS still cost a fortune while Windows and UNIX's (esp BSD's) were "good enough." That is, DEC & Compaq seemed to ignore the lessons of Worse is Better essay about the market power of an imperfect, mostly-working solution. They might have gotten further lowering their price alone. Maybe still have to target commodity hardware, though.