October 4, 2000

Linux overload

Author: JT Smith

By Jack Bryar
NewsForge Columnist

Open Source business

Out sick for just over a week and I get flooded with stories about Linux
forking, Microsoft tanking, and all sorts of new Open Source adopters coming out of
the woodwork. Fresh from a fun-filled visit with the staff of the Dartmouth
Hitchcock Clinic & Medical Center, my mind has been reeling this
week. I think it is due to a couple of causes. Certainly an insufficient
supply of oxygen to the brain, courtesy of a nasty case of
asthma-becoming-pneumonia, didn't help. Neither did a surfeit of Linux
news. I can't cope. My personal news collector pulled down more than 130
business stories involving Linux or Open Source just on Friday while I
was visiting with Dartmouth's resident pulmonologist, my new best
friend. One hundred thirty stories! Even properly oxygenated, my mind can't process
that much stuff.

Even in my impaired condition, I saw several themes in all this
stuff. Among them:

Is it just me, or did every major IT vendor left
standing on the sidelines come up with a Linux play over the last week?

Based on what I saw, it looked like everybody got into the
act. Cisco, a compulsive buyer of new communications technology, spent
$369 million to buy both Vovida Networks and IPCell Technologies. Vovida
is one of those Open Source companies that have emerged in the telecom
sector. Another telecom company, NTT Communications Corp., is often
the "early adopter" of new technology in Japan. I pay attention
to its Open Source announcements. NTT said it would spec a TurboLinux
-based server as the core of its e-business infrastructure. NTT also
said it would join several other Japanese companies investing in a new
VA Linux subsidiary in which Sumitomo will be the junior partner.
Toshiba Engineering may pitch in with on-site service and help-desk
support for the new VA Linux Systems Japan. Elsewhere, Toshiba and IBM
Japan announced that they would co-market and sell Linux-based computer
systems featuring IBM servers and Toshiba's Linux-based cluster
software. NTT announced they were going to buy the packaged systems.
This doesn't even count the news about Informix and Open Source ... or
the Motorola announcement ... or the Samsung/Red Hat announcement ...
or ...

In certain applications, Open Source's computing lead is now
considered insurmountable.

For many of you this is old news. But I've got a fistful of recent
magazines and newsletters claiming that Microsoft has been effectively
shut out of important applications. The Web server market is gone, they
say. They suggest that the Apache web server is so feature-rich, has
such a technical lead, is supported by so many auxiliary projects
(Apache XML, The Jakarta Project, etc.), and is supported across so many
platforms, that as Network Computing
recently said, "It'll be a cold day in hell before Microsoft or any
other commercial Web server vendor can [compete]."

Likewise, high-end environments are being dominated by cluster
servers, especially in scientific applications, where Linux is now the
default OS. Microsoft's NT architecture should lend itself naturally to
clusters. The company has supported development at Cornell, U.C. San Diego
and the University of Illinois. Yet, Open Source Beowulf systems are
flourishing; the academic and scientific community continues to
contribute additional development resources, and the antagonism against
Microsoft has grown so strong in these quarters that several newsletters
suggest that Microsoft should simply cut its losses. If Microsoft took
the advice, it would mean walking away from bleeding-edge computing
applications in telecom, robotics, neuroscience, knowledge management,
meteorology, and biotech.

So ... why was so much controversy generated this week about
whether Open Source makes sense as a business or development
model?

Computer Reseller's News carried a controversial interview with MIT's
uber-economist Lester Thurow, who declared that Linux and Open Source
"can't possibly work. It's open architecture. People change it, and the
changes aren't compatible. The only way software can be compatible is if
one company owns it." The responses were predictable. Many pointed out
that proprietary vendors have a history of making changes to their
platforms and stranding their developers and partners in the process.
Others pointed out that open, multi-vendor "standards" have been
critical to rapid technical development in a variety of other
fields.

But some advocates worry that Thurow might be right. Linux's biggest
challenge is coming from too much growth rather than too little. That's
becoming a big problem as widespread adoption has been accompanied by
multiple patches and modifications.

Is Linux going to fork?

Some Open Source advocates are saying it's
inevitable. Others think it has already happened. Is it Linus Torvald's
fault? Or are Open Source applications too diversified and growing too
fast?

I'm a recent subscriber to a Linux Kernel mailing list. I've
been surprised by the number of participants who seem exasperated by
Torvald's attitude toward his own OS. The complaints surfaced on
Friday's Radio Wall Street. One commentator complained that Linus was
too interested in a system that was "fun" and didn't care about the
needs of enterprise systems developers. Others have noted that
enterprise equipment and systems developers have begun to spec
exclusively around patched versions of Linux, especially those coming
from Red Hat and Mandrake, to meet the needs of their clients. When
does a patch become a fork?

And when do the other flavors of Unix effectively become forks of
Linux? Maybe it's my illness, but the more I read about the embrace of
Open Source by the big hardware vendors, the more concerned I get. For
example, IBM has been on a mission to port most of the major GNU and
open source apps to the AIX kernel. Is this an effort to support Open
Source, or to position its Unix kernel as a more robust and better managed
alternative to Linux? (Or both?)

Finally, I'm not sure if some emerging flavors of Embedded Linux
can even really be considered forks, or even Linux. I know that
something that looks "a lot like Linux" has taken hold in embedded
computing systems. I read the news coming out of the Embedded Systems
Conference in San Jose last week. A Beacon Research study claimed that
embedded Linux development jumped 1400% in the last year. As many as one
third of embedded systems developers have plans in place to employ a
Linux-based platform.

Open Source makes sense in embedded systems. Margins are small.
Developers want to avoid royalty payments. Open Source environments are
easier to service. But John Fogelin of Wind River Systems cautions,
"The embedded market is inherently fragmented, and therefore does not
lend itself to ... community-based open-source development." Fogelin
suggests that the lack of common platforms and functions make it hard to
adhere even to basic standards. As the kernel is patched or
simplified, are these variations forks? Or even Linux? And does it
matter?

Word comes from several sources that Intel is going to be swamped
with Gigahertz chips next year.

If compelling applications don't start to emerge -- fast -- Intel is
in trouble. The company has dumped a big chunk of its $4 billion
research budget into software development, not so much to create
products as to seed development in power-hungry applications like voice
and video recognition and MPEG-4 video compression. And it's been contributing
key data security elements
to the Open Source community in an effort
to drive serious e-commerce applications.

Increasingly, Intel is looking to Open Source community to speed
development. It's been no secret that Intel has become frustrated by
Microsoft's sluggish development pace. No one thinks the Open Source
community will solve the company's short-term problems. But strategists
like Harry Edwards of Avcom Technologies suggest that Intel will have
to depend on a "Lintel" alliance with the Open Source community if it is
to drive demand.

The boys from Redmond are back to their old tricks.

Perhaps it is to celebrate the fact that David Boies won't be
around at the Justice Department to pick on them any more. Perhaps it's
because Microsoft has its antitrust case safely buried in a Court of
Appeals.
Independent innovators are being targeted, and lawsuits are flying, but
the stock market doesn't believe in Microsoft's inevitability any more. Too
big? Too awkward? Bad business model? (Bad leadership?)

But there's still plenty of joy in Redmond. There's a good chance Judge Thomas
Penfield Jackson's ruling will be either overturned or sharply limited.
There's an even better chance that the case will be hung up for the next
couple of years.

It may be a coincidence, but Microsoft spent last week hassling old
antagonist Ray Noorda, formerly of Novell. One of Noorda's start-ups,
Timpanogas Research, says Microsoft is suing it because the company is
trying to help develop Linux development tools. Microsoft says that when
Timpanogas illustrated how to configure NT to interoperate with Linux,
it poached Microsoft's intellectual property. Really? Not long ago,
Microsoft did nothing when the NT F/S was ported over to DOS by a third
party. Once intellectual property is left undefended, it's usually hard
to claim infringement later on.

Microsoft is also messing with bigger companies that can fight back.
Sony is reportedly unhappy over Microsoft's X-Box gaming system. Both
companies have ambitions for the home market and most observers expect a
titanic clash as home PCs and game systems begin to merge. But Sony has
a strong installed base and plenty of dedicated developers. Microsoft
risks having a real fight on its hands.

Many analysts think the company is picking fights with too many
companies, is trying to manage too many projects and is doing a bad job
with nearly all of them. Microsoft's market valuation is going into the
tank. For once the investment community has got it right.

One of the toughest analyses was written by Stu Johnston, a guy who actually
understands computer software. Johnston has written about Microsoft for
a dozen years. He's no basher. This is a guy who just loves Windows 2000. So his recent article in Information Week was a revelation.
Johnston's prose may have lacked a set of consistent analogies (he
compared Microsoft to "a chameleon," an "amoeba" and a "velvet
sweatshop" within a couple of paragraphs) but his premise was consistent
and devastating. He said the X-Box is an attempt by Microsoft to build
its own PC, infuriating hardware manufacturers. He said that Windows
Everywhere is dead, that Windows.Net might as well be, and that Windows
CE is forking into a half dozen OSes that no one wants. He said that the
company is so far behind Open Source competitors in embedded technology
and the Internet that there's no possibility of the company catching
up.

Johnston joins many tech and business writers who suggest that the
company has finally grown too big and unwieldy for its own good, and
that it's time to break it up. One advantage to a breakup; a Microsoft split into a half
dozen companies would mean that at least five of them wouldn't be run by
Bill Gates. Microsoft's publicity machine continues to promote the
borg-master as the geeky genius of IT, but his potential workforce isn't
buying the image. Last summer I contacted IT department heads at better
than a dozen colleges who said student interest in a Microsoft career
had never been so low. Johnston says interest is nearly as low among the
company's star employees, who are heading for the exits in droves.

If
Microsoft survives its antitrust case it may encounter something worse --
real competition without the IT superstars it needs to fight back. I
just can't tell if the competition will come from a unified Linux
community or a fractured set of Unix/Linux and embedded systems
developers.

Category:

  • News
Click Here!