In-Reply-To: <199507061324.VAA28052@vector.wantree.com.au>
Message-ID: 
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII


On Thu, 6 Jul 1995 nathan@wantree.com.au wrote:

> I get the impression that Jeremy and Rex are arguing the same point - from 
> different ends.
> 
> This is less an argument of Unix vs Windows as it is Server vs Client.  The 
> Net was created and supported initially by the powerful Unix/TCP/IP servers. 

More importantly, Unix was a product of the INTERNET.  When dealing with 
the broad range of economic and technological forces present in a 
community of thousands of VARs, OEMs, and Consultants, the need for 
common API's driven by infrastructures which provided source code, and 
could be ported to hundreds of diverse platforms, became a major driving 
factor in the convergence of the internet and Unix.  Quite simply, Unix 
(As we know it today) is a result of the Internet, the internet is a 
result of Unix.

Anyone who remembers the original Version 6 Unix, released to educational 
and research institutions, would barely recognize it when compared to the 
technology of today (Mach, OSF/1, Linux,...).  Almost everything has 
changed "under the hood".

Unix was an object oriented operating system before anyone even knew what 
"object oriented" was, look at the device drivers and you can see what 
spawned what.

Threads, OS/9 had them back in 1978.  Also DLLs, Shared Memory, and IPC.
Unix adopted them in 4.3 BSD and SysVR3.  In fact, almost every major 
innovation in software engineering has either originated from UNIX or
has been incorporated.

The biggest threat to Unix was the prospect of the SmallTalk operating 
system.  Unfortunately, it became a political landmine when Mac started 
"taking posession" of most of the PARC concepts and practices.  Still, 
most of those concepts had already been incorporated even before the
MacIntosh came out.

If Microsoft put out the Source Code to Windows-NT on a "play or pay" 
basis, they could take advantage of the 50,000 staff-years that are put 
into Unix Enhancements every year.  Of course, Microsoft might have to 
pay out a bit of that 12 Billion to the nice people who contributed 
without knowing.

>  Yet as most people don't and never will have Unix on their desktop, the 
> growth in the Net is due to those programs and clients being ported to their 
> desktop of choice - which is more often than not Windows.

You assume that people will never have Unix on their desktop.  I contend 
that the "Quantum Leap" from Windows 3.1 (2meg ram, 40 meg drive, 386) to
Windows NT/95 (32 meg ram, 1 gig drive, Pentium 90) has created the same 
type of opening that was created when IBM introduced Microchannel, along 
with the "death" of ISA.  The vendors revolted, the customers revolted, 
and the retailers revolted (after being stuck with $$$ of unsold MC periphs).

Unix (especially SlackWare Linux) fills a wide-open niche for a "low-end" 
32 bit operating system.  BSDI and SCO are competing for the "low-end" 
business market, especially in the small-business, franchise, and branch 
store domains.

The Windows "Bundle" has left very few developers of third party 
applications willing to invest large amounts of money into NT/95 
support.  It's highly likely they will do little more than recompile the 
16 bit code for 32 bit mode.  Many applications vendors will be offering 
CDs containing Windows, Mac, Linux, and BSDI/SCO.  Some of the video card 
manufacturers are including the 3-D features required to support PHIGS/PEX.
(A powerful 3D language used to produce "Real-Time" 3d) - a Unix thing.

The Internet/Unix relationship goes very deep.  People have been 
marketing applications, hardware, and services for Unix on the internet 
for years, through public distribution of "subsets".  The Interviews
toolkit comes in "command line support" mode, but is sold with dialogue 
editors, UIMs, dozens of button-caps (graphical buttons), and enhanced 
tools that are compatible with the original, but reduce the development 
times.  Since the target was corporate america, it was easy to give away 
samples and sell the "site license".

Selling to an individual market precipitated the need for functions like 
Digicash and First Virtual.

> The point is that it really is not a matter of numbers of people using the 
> flavour of OS that is important.  By definition there will always be many 
> more clients than servers, and due to the specific nature of both types, they 
> are unlikely to be the same platform (although who knows in five years time 
> ..)

In all likelihood, there will be further evolution and convergence.  The
Unix vendors will continue to make their products much easier to use. 
Microsoft is already selling MFCs for SCO and BSDI.  Gates already
personally owns about 20-30% of SCO and probably has access to a sizable
chunk of BSDI.  For obvious reasons, Microsoft is not interested in
paying AT&T any more than it has to.  Supporting Linux would force USL to 
drop it's royalty and price "floor" as well.

The most frustrating part of it is that the biggest threats (BSDI, 
SlackWare, Ydraggisil) are all closely held with little desire to go public.
The fact that Linux is distributed with source, and that BSDI sells their 
source code for under $1000 (to all but the "critical" AT&T code)  
reduces the support requirements substantially.  Many stores will install 
Linux for an extra $300.

Hardware Vendors are actually writing and contributing their own device 
drivers.  In some cases, they have to be downloaded via the internet.

> Suffice to say that the Web would not exist without Unix and that it is 
> unlikely to have become the growing force for good it is now without Windows. 
>  Who is to say which is the most important?

The real question is: "Can 12 billion dollars compete with 5 Million 
Man-hours (worth over 200 billion)"

> If I can see further it is because I stood on the shoulders of giants.

Until Internet/Unix programmers stood on the toes of other programmers.
Can Microsoft continue to retain tight control over it's intellectual 
properties and still compete against it's own customer base.

In each market where Unix has prevailed, the break-through point was when 
the cost of Software (Purchase, Maintainence, Training, Upgrades...) 
exceeded the cost of the software on which it ran.  Windows NT - 
unlimited user (required for internet servers) costs several times the 
cost of it's underlying hardware.  Windows-95 costs several times the 
price of it's underlying hardware.

Microsoft will be confronted with a worthy competitor against which it is 
as poorly armed as was IBM against the PC and Unix powered Minis.

> 'Nuff said.

	Rex Ballard
	Standard & Poor's/McGraw-Hill
	Opinions expressed do not necessarily reflect
	the Management of the McGraw-Hill Companies.





From rballard@cnj.digex.net Thu Jul 13 23:46:39 1995
Status: O
X-Status: