Message 00398 [Homepage] [Navigation]
Thread: oxenT00397 Message: 2/2 L1 [In index]
[First in Thread] [Last in Thread] [Date Next] [Date Prev]
[Next in Thread] [Prev in Thread] [Next Thread] [Prev Thread]

[ox-en] Amy Wohl: operating systems are becoming less important



AMY D. WOHL'S OPINIONS - Volume 2, Issue 8
February 22, 2002
http://www.wohl.com/issue.htm

OPERATING SYSTEMS WARS ENTER NEW PHASE

It's interesting how seemingly unrelated things sometimes occur in such
close proximity that you can't help noticing their underlying
relationship.

In the last few weeks, I've noticed a number of events related to
operating systems, a topic I don't usually spend a great deal of time on
(although I confess to a continuing fascination with the market's
increasing passion for Linux).

Let me see if I can list them out and then try to relate them for you.

(1) I went to Sun's Analysts Conference (as you know from last
week's newsletter) where Sun announced it's new, deeper interest in
Linux.

(2) At that same Analysts Conference, Sun CTO Greg Papadopoulos
announced a new "meta operating system" called N1.

(3) In my University of Pennsylvania class on Why Technologies
Succeed or Fail this week, I made a formal comparison between IBM's OS/2
and Microsoft's Windows NT, leading to a more general discussion on the
state of the operating system market today.

(4) In one of the 50-75 newsletters, weblogs, and e-lists I get
every day, I received some information recently about Dataquest's 2001
server market share numbers (the usual tease, designed to entice you to
buy their report).  Interestingly, IBM and Sun together accounted for
just over half of all of the $16.7 Billion in servers sold (of all
types).  But Sun's market share has dropped to 21.4% and IBM's has gone
up to 29.3% (of all servers, based on revenue, not units).  Intel and
Unix Servers make up $14.4 Billion of that $16.7 Billion market (but
remember that Linux as well as Windows might be running on Intel
servers).  The remaining $2.3 Billion is mainframes, minicomputers like
IBM's i-series (the former AS/400), and a relatively small number of
Apple servers.

It's hard to figure out how much of that is Linux - IDC estimates that
about 27% of all servers use Linux but that includes desktops used as
servers and existing servers repurposed by their owners with free or
purchased Linux software.  (Gartner argues the number is much lower,
probably in the area of 6 to 8 %.)

What does all this mean?

OPERATING SYSTEMS ARE BECOMING LESS RELEVANT

Today it's becoming less important what operating system is running on
your device (desktop, handheld, whatever) and more important that you be
able to easily run a standard browser.  This is because the most
frequently used applications are email and web access, not personal
productivity applications that are operating system dependent.

Also, developers are often writing not to an operating system API, but
rather to an application platform (a web server perhaps, important
application like Office, SAP, etc., etc.) or a development environment
like web services, e.g., .NET or J2EE.

With handheld devices, operating systems are even less relevant.  This
year, analysts believe devices with embedded operating systems
(handhelds, telephones, cars, intelligent refrigerators, and personal
medical equipment) will exceed personal computers with traditional
operating systems.  Most people neither know nor care what the operating
system on their phone is; in fact, very few people even know phones have
operating systems.

While embedded versions of Windows have been more successful recently,
generally other products, optimized for small devices, such as embedded
Linux, PalmOS, Symbian, and Wind River have been the operating systems
of choice for these environments.  In these small environments, small
footprint and appropriate performance have so far outweighed being able
to match desktop operating systems and access existing applications and
developer communities.

If Sun, which has long argued its superiority in offering a single
operating system, Solaris/Unix, "from a less than $1,000 system to a $10
Million system," can see the need to provide a broader Linux offering,
we think the market is speaking, demanding more choices.  We believe
that IBM's commitment to offering Linux across its entire product line,
and the interest of other hardware vendors in providing Linux choices
helped to create an environment in which this became an obvious move.

It is too soon to see if this point of view will prevail.  Customers
don't buy operating systems per se, but rather solutions that provide
robust environments and access to lots of software and development
skills.  If Linux can offer that, it may create a place for itself in
the enterprise market.  If it doesn't, that will be because it failed to
pass the ultimate test - provide useful, reliable function to customers
at appropriate prices.

NEW KINDS OF OPERATING SYSTEMS MAY BECOME IMPORTANT

On the other hand, we may want (and need) new kinds of help in managing
increasingly complex computing systems.

Sun's N1, which is a kind of operating system to provide integration of
diverse systems elements for a network-based, distributed computing
environment, knowledge of their state, and control management seems, in
some sense, like a modern version of a mainframe operating system.  That
is, it performs some of the same tasks, but it performs them in a very
distributed environment and it may ultimately be able to integrate not
only Sun's components - servers, storage, and software - but also
heterogeneous components from other vendors.

At the same time, IBM is exploring the idea of e-Liza, a series of
hardware and software improvements to its servers and software which
will allow systems to be self-diagnostic and self-managing (sound
familiar?).  Even more ambitious is their Autonomic Computing project
http://www.research.ibm.com/autonomic which will try to make
increasingly complex systems simpler by treating the entire computing
environment as if it were a human body, with the autonomic (or
automatic) nervous system making adjustments to the environment and
managing the system.

Such advanced meta-operating systems may, in the final analysis be much
more interesting and important than the operating systems of today since
they will buffer both users and systems operators from the complexity of
the computer and its transactions and tasks and simply permit them to
USE the computing environment without the need to learn arcane commands
or rules.  That couldn't happen soon enough!


_______________________
http://www.oekonux.org/


Thread: oxenT00397 Message: 2/2 L1 [In index]
Message 00398 [Homepage] [Navigation]