In-Reply-To: <4rafg2$k40@digital.netvoyage.net>
Message-ID: 
References: <833058917.18622.0@melech.demon.co.uk> <4r3s3b$r2r@news2.inlink.com> <4r4v5b$he7@digital.netvoyage.net> <4r8mea$as7@news2.inlink.com> <4rafg2$k40@digital.netvoyage.net>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII


On 1 Jul 1996, Joe Sloan wrote:

> In article <4r8mea$as7@news2.inlink.com>, Sangria  wrote:
> >In article <4r4v5b$he7@digital.netvoyage.net>, jjs@digital.netvoyage.net says...
> >
> >www.sequent.com
> >
> >They have a bunch of white paper sitting on their web site along with a
> >link to some demos.
> 
> I looked at the site, but they seemed to be a hardware house
> which isn't too religious about what OS is used as long as you 
> buy their boxes. (Like DEC ?) I guess what I'm saying is that 
> they don't seem to be some sort of flaming microsoft cheerleaders,
> but are merely keeping all their options open -

Sequent makes a very nice fault tolerant version of Unix.  It's hard to
tell the difference unless you are porting a program like "top" that
probes into unix kernel values from the kmem and vmunix tags (what are
you doing there in the first place you rotten little kid :-).

> >The difference there is that these aren't the same flavors of Unix.
> >Nor is the migration if you wanted to "scale" up a trivial one.  Going
> >from a Unix capable of running on a lowly PC class machine (a 486 for 
> >example) to something that can serve and entire enterprise would be a
> >neat trick.  

I've used Ultrix, AIX, SunOS, Solaris, and HP OSF/1, also AT&T version 7,
BSD 4.2 to 4.4, SCO, and Interactive.  There are subtle and not so subtle
differences (HP's printer management still makes me nuts), but most of it
is source code and shell script compatibile.  In addition, most of the
Rapid Application Development languages like Perl, Awk, Sed, Lex, Yacc,
and TK/TCL hide the really ugly stuff very well.

The most frustrating thing is when I am using a "supported" Unix
implementation and discover that I don't have a tool or language that I
have been using on Linux.  Sometimes I can get it on a "Complimentary
Software" (Like complimentary Prime Rib with your glass of wine :-) CD.

When worst comes to worst, I can get the source from the Linux CD and
compile it under gcc on the target system.  Once in a while I have to
deal with some piece of code that thinks that he can byte copy length
values from a string buffer to an integer without using htons, but such
instances are increasingly rare.

> Linux is getting there faster than most people thought -
> Name one other OS that runs just fine on a lowly 386-sx, a DEC Alpha, 
> a Sun Sparcstation or a Fujitsu Supercomputer - Hint: It isn't 
> windows "new technology"!

I was actually suprised to see Linux ports for platforms that were
traditionally Unix based.  On the one hand, it seems like there is little
to gain.  On the other hand, it's the closest the Unix community has ever
come to full source code compatibility.

Given the capacity of a CD-ROM and the amount of "common" information such
as help's, fonts, info pages, scripts, and other data files, it doesn't
take a rocket scientist to realize that putting several hardware specific
ports on a distribution CD makes sense.  You could also always get your
hardware specific binaries off of the internet.

> >BTW, who's Todd?
> You don't know Todd [Microsoft]? Where have you been? This manager 
> from microsoft was bombarding the Linux discussion group with daily
> rations of clever, amusing microsoft sermons. He was involved in half
> a dozen threads, arguing about the definition of "multiuser" and
> other trivia, lacing his arguments with winsome, self-effacing humor 
> and cleverly twisted arguments - what a master of public relations...

Hey, Sun's been doing it for 10 years, it's worked well for them.  With
any luck, Todd is getting a hard lesson in what it will take for NT to be
accepted (It still has a chance, but I don't think Microsoft will be
willing to do what it takes).  Besides, Microsoft is now getting most of
it's revenue from application sales and services.  It may turn out that
Microsoft will slap a Windows-95 "look and feel" on Linux and develop
it's own "add-on package" for this platform.  This is what DEC, IBM, and
HP/Apollo did.  Microsoft has spent almost $4 billion promoting Windows NT
and subsequently Windows-95, and has barely broken 25 Million users (less
than 10% of it's own market).

The users want to upgrade to something, and are progressively less
concerned about backward compatibility (Note push to NT vs. 95, given NT
is less effective at running legacy applications).  There are also lots of
386 and 486 computers, along with 4-8 MB Pentiums that are too weak to run
NT and would thrive under Linux.  Short of Microsoft offering to buy
old PCs from users for the purpose of destroying them, there is too much
of a void to be filled.

The precedent is similar to the period when IBM decided to support only
Micro-Channel on OS/2.  Users of ISA boxes went with Windows 3.0.  By the
time IBM realized that they had seriously damaged their reputation wtih
the user community, it was too late, Microsoft became the leader of MIS
technology and IBM became the "Orphan Machine".  Another precedent is when
IBM decided that all of their users would upgrade to MVS 4.0 and purchase
all of the mandatory application upgrades to support it.  MIS directors
confronted with a proposal to spend $3 million in software upgrades to a
system that couldn't even manage ASCII files and requests for $25,000 to
$250,000 for Unix based systems that could run Oracle or Sybase at speeds
approaching the fastest IBM mainframe's DB/2 performance decided it was
worth the risk to try these Unix prototypes.  MVS 4.0 was announced in mid
1991, by December 1992, the stock was in a free-fall as expected orders
were never received and as previous orders were cancelled (because the
Unix systems were outperforming expectations).  By 1993, IBM was frantic
to have a full-featured SMP implementation of OSF/1 which was "compatible
with Sun".

As Word Perfect, Perfect Office, Lotus SmartSuite, and other "Full
Featured" office automation packages start offering "3.1, 95, and Linux"
versions, they will discover that Linux is a very receptive market.  They
will also begin to take advantage of multitasking (vs multithreading) and
create "data-stream" applications such as spreadsheets that can have cells
or rows fed from a "pipe command".  The paradyme will be more like Sun Net
Manager, but the applications will be things like simultanious tracking of
industry groups containing 100-300 individual stocks.

> >>Like I said, windows "nt" is a decent pc server, but anyone who
> >>beleives all the hype about it being some great breakthrough
> >>probably beleives deep in his heart that the earth is flat too...
> >
> >Are you just peeved that NT does scale beyond a lowly Proliant?

Windows NT is a "skeletal" server.  It has much of the minimum
requirements to make it function in trivial applications such as simple
file servers, sql servers, and web servers, but high performance depends
heavily on multithreading which does not provide the protection and
stability required for more complex applications such as real-time
distribution, e-mail, news, statistical analysis, and server sharing.

> Oh, please...
> 
> If the wealthiest man on earth, running the one of the wealthiest 
> corporations on earth, couldn't manage to buy into a high-end port,
> that would be surprising. 

I found it interesting that Sun didn't come to the NT party.  Did they
refuse to play? Did Gates forget to invite them?  Either way, they are a
formidable opponant, as is the "Internet Panopoly" of Unix programmers,
users, and administrators.

> The thing of it is, Sang, that all the
> hulabaloo over windows "new technology" is driven by the vendor's 
> advertising,

It's always interesting to look at the "New Technology" that has been
announce by Microsoft, and the timing of those announcements.

	By the time Windows 3.0 had come out, X11 was in it's third
	release, a mandatory requirement of the Federal Government, and
	was rapidly gaining acceptance in the user community.

	By the time 3.1 was announced, people were beginning to specify
	X11/R4.  The Microsoft booth was almost desolate.  By the time
	Gates was to give his "Victory Speech", he realized he had to
	announce a product that could be a credible competitor to SUN.
	He proceeded to describe this "New Technology Operating System"
	that would match the description of the existing technology SUN
	was demonstrating at the time.  One reader actually remarked "he
	must have spent the entire night playing with that SUN box to
	come up with that Spec".  Gates promised a "Better Unix than
	Unix".

	For almost 3 years, the industry waited for the new "Miracle
	System" to appear.  When it was finally released, bugs and all,
	the MIS managers who had bet their careers on NT by holding off
	Unix purchases and upgrades were suddenly developing peptic ulcers
	as NT workstations collapsed.  Particularly distressing was that
	NT operating system required more expensive hardware than a Sun
	Sparc 5, and didn't perform as well.

	In what was supposed to be his NT victory speech, Gates promised
	an unresponsive audience an leaner operating system code-named
	"Chicago".  MIS managers went home grinning and promising that
	"Chicago" would be the cure-all.  As NT sales fell flat (50% of
	the product in the field was "give-aways") and 95 ran later and
	later, MIS managers who had bet their futures on Microsoft were
	starting to face the loan-sharks.  Unix proponants were bringing
	in CD versions of Linux, upgrading PCs in "Dual Boot"
	configurations, and setting up Web browsers on their boss's PCs
	which were served by the 486's that were shelved because they
	couldn't support NT or 95.  Machines that crawled under 95 demo
	software seemed to dance under Linux.  Pentiums actually seemed
	like overkill to these little monsters.

	Soon, proponants of Unix were starting to look very good, able to
	develop prototypes on Linux and upscale them to politically
	correct machines like HP or Sun.  Of course, these companies
	weren't stupid either.  They started producing big ticket machines
	like the Sparc/2000 with 64 processor cages and FDDI clustering
	capabilities.  Metaservers became known as fire-walls and could
	make a complex cluster of SparcServers and NFS servers look like a
	single machine.

	More than one MIS manager has become a shining star by delivering
	"the moon" on a humble little Linux box, and then given the
	go-ahead to implement a much larger implementation.  Linux has
	become a powerful marketing tool for Sun, Hewlett Packard, and
	IBM.  It gets people away from the Microsoft muse.

	Meanwhile, back at the ranch.  Linux has been popping up all over
	the place as BBS operators replace their WildCat systems with
	Linux and get frame-relay connected to the internet.  In very
	little time, the majority of the servers are Unix Based.  In
	addition, there is a growing acceptance of Internet tools such as
	netscape browsers, Pegasus and Eudora E-Mail, and there is
	a proliferation of TCP/IP and NFS "Internet Kits" which work with
	high-end Winsock packages.  Suddenly, CIOs and CEOs want
	Intranet, internal networks that function exactly the same way
	internet tools work but behind the fire-wall.
	
> whereas the universal acceptance of Unix is a result
> of the excellent design which led to such popularity in the
> first place.

	Even more important than superior technical design is the ability
	to rapidly produce extraordinary results.  A 20 line perl script
	pipelined to a gplot can turn boring statistics into information
	which can be used to justify purchasing decisions.

> Mark my words: windows on Sequent will never be anything more than a
> curiosity, and will never capture any significant market share.

	DEC is selling a lot of Alphas to people who can't tolerate the
	delays and sluggishness of NT on the Pentium.  Of course, they
	can get even better performance under Digital Unix or Linux, and
	a "loaner" SCSI drive and a CD is all it takes to prove it.

	MIS Managers who have bet their careers on NT have been getting
	almost fanatical about their position.  In many cases, they are
	even firing people for putting Linux on their workstations.  They
	put out edicts which demand that ONLY approved software (Microsoft
	Office, Lotus Notes, Visio, and Microsoft products) will be
	allowed on the workstations.  
	
	The first justification is copyright
	management - which creates a problem since Linux encourages
	replication.  Then they claim security issues - until the Linux
	user creates a workstation that is so secure that even the Network
	Adnministrators can't get in (Linux is easily configured as a
	fire-wall).  
	
	Finally, they ask the Unix proponant to come up
	with a bid to implement a project - knowing he will estimate
	it based on Unix development time - and demand that he implement
	it in that time-frame on an NT (after sending the bid to upper
	management of course).  The Unix proponant must then either face
	dismissal for insubordination or dismissal for failure to meet his
	deadlines effectively.

	Microsoft is actually coaching some managers in how to deal with
	difficult Unix people.

> OTOH I predict that the popularity of Linux will continue to grow at a 
> brisk rate, and Linux will enjoy wider commercial acceptance in the 
> near future.

	If Linux become the Workstation and "Departmental Server"
	workhorse, creating openings for other "Industrial Strength"
	(politically correct) UNIX Enterprise servers such as the
	SparcServer/2000 or the HP/9000, it will have done it's job.

	If Linux use on the workstations result in placing SGI Indios and
	SparcStation 20's into the contention for "Specialty 3-D and Video
	Composition" work, displacing NT on the desktop.  If Linux in the
	"Back-Office" results on more SCO or SUN on the trading floor, it
	will have broken the Microsoft Stranglehold on the Software
	Market.

> jjs
> 
> 

	Rex Ballard - Director of Electronic Distribution
	Standard & Poor's/McGraw-Hill
	Opinions expressed do not necessarily reflect
	the Management of the McGraw-Hill Companies.
	http://cnj.digex.net/~rballard

	Soon to be moving on (Target of a Microsoft Biggot).




From rballard@cnj.digex.net Wed Jul  3 21:15:07 1996
Status: O
X-Status: 
Newsgroups: comp.os.linux.advocacy,comp.os.ms-windows.nt.advocacy