Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!isis!rballard
From: rballard@isis.cs.du.edu (Rexford E Ballard)
Newsgroups: comp.arch
Subject: Re: Sun's Competitive Strategy
Summary: General Competitive Strategy
Keywords: sun, open systems, portability
Message-ID: <1990Dec1.030450.28687@isis.cs.du.edu>
Date: 1 Dec 90 03:04:50 GMT
Sender: Rex Ballard
Reply-To: rballard@isis.UUCP (Rexford E Ballard)
Organization: Great-West Life (Via nyx)
Lines: 70
Distribution:usa


Sun, or any other vendor for that matter, has a simple competitive
strategy.  Be "open" enough tbe the beneficiary of "commodity" support
software and peripherals, "unique" enough to warrant a higher mark-up, and
"closed" enough toensure growth and migration to new standards as they are
developed.

There have been many "cold steel" manufacturers of "standard" "cold iron"
boxes such as 68020 running BSD-4.2, CP/IP, and SCSI.  Sun survived by
adding valued features such as windows (SunView) back when X-10 was still
an experimental project and Macs came with 128K.  The migration to Xview
was somewhat traumatic at first, but as applications are ported to the
commodity environments (Common subset of Xview, Motif, ICCM), applications
previously only available on PCs are showing up on UNIX.


There have been many attempts to turn an MS-DOS PC into a more "unix-like"
box, with simple multitasking, windows, device independent graphics, and
interprocess communication.  Unfortunately, each attempt (including early
versions of windows) failed, primarily because of applications such as
Lotus 1-2-3 that literally took over the machine.  Each upgrade to IBM
graphics mandates an upgrade to Lotus.

When the IBM-PC first came out, IBM published quite a bit of technical
documentation.  In addition, they had a "hacker's contest", offering
incentives to those who could write softwarehat could take advantage of
the unique features of the PC.  The intent was to motivate businesses to
to switch away from CP/M and MP/M to the PC.

This "openness" was a double edged sword.  While it did render CP/M
obsolete, and it did prevent excessive cloning, it also made it difficult
to enhance the basic PC.




To be fair to Sun, they have adopted some very open standards including
SCSI, TCP/, X-11, and UNIX.  They also also provided SysV and BSD 4.2
support in the same machine.  They have also been good about defining and
opening their standards.  NFS, Yellow Pages (NIS), and RPC are examples of
standards defined, liscenced cheaply (note -ot free), and well documented
through well known sources (Internet RFCs).

It would be nice if Sun would ease up on the "Smoke and Mirrors" that make
Xview applications seem incompatible with other environments.  It would be
nice if they could "play nice" with other window managers (XView
initialization is sooooo slooooow).

On the other hand, look at the trouble they have had to go through to
maintain backward compatibility with their previous window systems.  XView
still proves support for SunView and NeWS.  This is real important when
you have to wait for 3rd party vendors to publish and release their XView
compatible software.  Unfortunately, there was so much hype over the
Motif vs Xview that vendors had to "wait and see".  It was interesting
that as the smoke cleared, there was an equivalent to "STDIO" that worked
on both platforms, along with many others.

I remember when MIT students passed around copies of the UNIX version 6
source code for PDP11/23s.  Back then the entire kernal was 128K and
were swapped by the context switcher.  You could hang 30 users on a
machine with less power than a 8Mhz PC/XT.


The original AT&T UNIX was almost public domain because antitrust laws
prevented them from selling the OS.  BSD improved it (to 4 Meg),
commercialized it, and liscenced it back to AT&T for SysV.3+.

The irony here is that a 3rd party vendor will write software to
the "lowest common denominator" (BSD, Curses, Plot, XWidgets, EPS...) to
be sure that he can "catch the next wave".'