Post by Scott Dorsey
And the first major difference was that the vax was actually a very fast
machine, especially at the price point, and the HP/3000 was really kind of
clunky. It was really a stack machine designed by instrument controls people
who thought IEEE-488 would be a great disk drive interface... which it might
be for laboratory controls but not for a database machine. The move to PA
helped but it was too little too late.
The HP 3000 suffered from "second system syndrome" from people who had mostly
not designed the first system. The HP 21xx series CPUs (packaged as the 2000 series computer systems) were originally developed by another company which HP
then purchased. HP was torn in 2 directions, with one part of the company want-
ing to make the 21xx an instrument controller and another part wanting to sell
2000s as timesharing systems. The second was wildly successful, the first not
so much. But much of the original CPU design and initial software (assembler,
loader, etc.) came from outside HP.
HP decided to build a 32-bit replacement for the 2000, using mainly people
brought in from outside HP (Burroughs and a few IBMers). That project quickly
spiralled out of control and was cancelled.
The 3000 came out of the ashes of that project (hence my comment about the 2nd
system effect from people who never actually finished a first system, either
the 32-bit system which was cancelled early in development, or the 2000 where
they just put the finishing touches on another company's design).
The 3000 promised 32 interactive terminals, batch, and real-time support simul-
taneously, all delivered at initial product shipment in 128KW. Needless to say,
that didn't happen.
I actually saw one of the original 3000's in the early 70's. It was quite a
sight to behold - aside from the CPU cabinetry there was a wide desk with 2
terminals, one for the operator and one for maintenance, with large panels full
of LEDs with stenciled and varnished designations. It was mainly serving as a
space heater at the timesharing service where it was installed. HP "loaned"
that company a pair of HP 2000 timesharing systems and upgraded another system
that was already on-site (the site had 4 HP 2000 systems, but one was older)
until the bugs could get worked out of the 3000.
Working the bugs out eventually meant recalling all of the systems from the
field, de-committing major parts of the promised functionality, and a delay of
several years until the CX series was released. Things gradually got better and
faster with subsequent software and hardware, and HP eventually transitioned
the environment to 32 bit hardware and beyond.
While all of this was happening, DEC had a huge experienced group of hardware
and software developers and had released many different families of CPUs, often
with multiple operating systems for each. So they were in a much better position
to release the VAX and VMS, although VMS didn't really hit its stride until
V4.0. Whatever you do, don't type Control-Y on a VMS 1.x or 2.x system - more
than just your process will go away!