What differentiates a Workstation from a PC (Re: What ...)

[Bennett Todd]

Warning: inflammatory opinions follow!

In article <1168 at mitisft.Convergent.COM>, kemnitz at mitisft (Gregory Kemnitz) writes:
>Software for personal computers (MS-DOS machines, Macs, Amigas) tends to cost
>generally less than one thousand dollars for all but the most super-duper
>special purpose software.  However, virtually everything for 'workstations'
>is atrociously expensive in comparison, if the software exists at all.
>[...]
>There is almost no general-purpose (non-techie) software for workstations,
>and what little there is costs thousands.

and so on. I think he hit the nail on the head, so to speak. For the
most part, workstations I've seen run PD or otherwise freely available
applications (obviously there are exceptions, but in terms of user-hours
running applications, or megabytes of disk allocated, I think the free
stuff dominates). By contrast, I would think that proprietary commercial
software tends to dominate the PC market. Applications for PCs seem
geared to being accessible to someone without forcing them to read
extensive documentation; workstation software is often well-neigh
unusable without reading the manual, or getting a wizard to show you
around.

Basically, I think what it comes down to isn't hardware characteristics
at all (the Mac II has just about all the hardware that would be needed
to make it a workstation, but you would never confuse it with one) but
rather the intended audience. PCs are targeted at folks who won't read
documentation, and who will be running commercial software. For doing
routine word processing and light-duty data processing stuff they are a
very cost-effective solution.

Workstations are targeted at development; either doing software
development for its own sake (CASE, anyone?) or doing other tasks which
are sufficiently specialized that you don't have a huge market to go
shop for commercial software, and so are going to be having to write it
yourself to get the job done. So, they need to be user-friendly, where
the first users are the programmers. PC vendors aren't excited about
UNIX; it provides access to a whole lot of flexibility, and it is terse.
These make it complicated for a neophyte to fire up the word processor,
and maintain their repository of documents. Those word processing
packages for workstations that succeed in concealing this sophistication
from the user do so at the expense of having to rewrite a bunch of
MacOS. Workstation vendors are more-or-less committed to UNIX (if they
want to sell any boxes in todays market!) because it provides access to
a whole lot of flexibility, and it is terse. These make it extremely
pleasant for experienced programmers to get their work done.

These days the predominant PCs seem to be the IBM family and clones,
which are running an evolutionary descendant of CP/M on an evolutionary
descendant of the 8080; the Mac, running an office-automation
application (instead of an operating system) basically developed by
Xerox; and the ST and Amiga, newer machines whose OSs contain influence
from both directions (in the user interface presented). You can
certainly develop software on any of these; however, folks doing big
stuff seem to prefer workstations, even though they cost more.

These days the predominant workstations all run UNIX.

You can pay the money for a workstation, pay a goodly whack more for
some spiffy software from Frame Technology, and enable it to perform a
job that a Mac II could handle for substantially less money. You can
also go the other direction, and buy a 386-based PC and jazz it up with
aftermarket UNIX, ethernet interface, mongo disks, and so on and make a
dandy workstation out of it.

I think the NeXT is an interesting case of a machine with a personality
conflict. It seems to be attempting to appeal to both markets, with
questionable success. It would need to be cheaper to compete as an
appliance; the overhead it pays in the attempt seems to make it less
attractive as a workstation.

There! Is there anyone out there I haven't insulted? No? O.K., I'll stop
now.:-)

-Bennett
bet at orion.mc.duke.edu

P.S. I *said* "Warning: inflammatory opinions follow!". Kindly email me
the flames; we don't need to clutter comp.unix.wizards with them.