Welcome to Open4Success

Marcel Kinard wrote:

 > Dale Pontius wrote:
 >
 >
 >> But it really is a Tax. It looks like a tax,
 >> walks like a tax, and quacks like a tax.
 >>
 >
 > Points granted. But when you compare the amount of the
 > one-time MS tax (~$150)

Industry average royalties paid to Microsoft average roughly
between $500 and $1000 per employee per year.  For corporate
agreements, this usually includes upgrades to Windows and
Office, and seat licenses to servers.  This was recently
increased by as much as 150% to between $1200 and $2500 per
employee for those companies wanting to participate in
Microsoft's new XP offering.  Many companies "panic purchased"
into the new plan, many other companies are finding that the
level 3 support and being double-billed for licenses (the OEM
pays for the license even when a corporate customer purchase
the hardware, then the corporate customer must pay for the
Microsoft supported licenses).  Some customers are looking for
options from the OEM, others are looking at alternatives such
as Linux, or just staying with current versions (and current
hardware).

 > to the recurring money spent by the CIO to
 >  support another operating system (C4EB, deployment, help
 > desk skills, applications, backup, license tracking,
 > server enablement, complexity, etc), the tax is so small
 > you don't care about it.

Both systems have hidden costs.  These costs are different, and
both sets of costs are important in calculating total TCO and
ROI.  The damage due to one virus alone (Melissa) was estimated
at over $4 billion - nearly $500 per person affected, this
included costs such as lost time, lost data, and cost of system
recovery.  Some virus, trojans, and worms can cost
substantially more.  Jane's resume, which reformatted hard
drives didn't spread as far, but the cost to those affected
often exceeded $5,000 per incident.

You also have service pack costs.  Most service packs and
upgrades are not well tested against third party software, and
seem to even contain code which deliberately damages 3rd party
software.  Windows 2000 had changes that prevented the use of
MQSeries, MQSI, and several other "competitive technologies".
Third party companies adjust, but customers are impacted by
delays and complications due to coping with these
incompatibilities.  Add roughly $2000/year for these costs.

You also have recurring problem costs.  Race conditions,
deadlocks, DLL interactions - also known as "DLL Hell", and
hard drive corruption (usually occurrs when the hard drive is
filled and files are removed to recover space), have no
permanent solutions, and result in loss of time and
productivity due to frequent reboots, not to mention
"reengineering" which includes reformatting the entire hard
drive, reinstalling all software, and recovering user data, as
often as 4 times per year with Windows NT.  This is why
Microsoft tried so hard to focuse on Reliability for Windows
2000.  They have improved the situation, but still are not as
reliable as most versions of Linux (yes there have been bad
releases, but they are the exception rather than the standard).
Add roughly $2000/year for these costs.

When you add it all up, you get real-world costs in terms of
actual expenses, lost productivity, and overall direct and
inderect costs that can exceed an average of $10,000 per
employee per year in some environments.  Generally, these costs
are made up by having the employee work overtime, which
increases turnover costs, which add more costs.

Linux has it's costs as well.  There are training costs, there
are costs for support, and there are costs related to
"tailoring", customizing your Linux environment to your
business (as opposed to the other way around), these can also
be significant, as high as $2000 per employee per year if you
were to focus only on those employees who have currently made
the transition.  Most of these employees, cover the majority of
these costs by taking their own initiative.  They tend to think
of their training and expertise in Linux as an investment,
which they can leverage to their own advantage.  Remember, the
earliest PC owners purchased their own PCs, brought them to
work, and used them to create reports and presentations that
impressed management.  This was back in the days when most
people either used a typwriter or a line-printer to print up a
memo which they would photocopy and put in everybody's inboxes.

Today, we live in a different corporate society.  We have
e-mail, web sites, chat/sametime, and deadlines that call for
quick and efficient communications.  Spending 3 hours picking
and tweaking the perfect format for a memo that could be
written in 30 seconds as a plain-text sametime request isn't
competitive.

Interconnection is also important.  Writing a document in a
proprietary file format, which must then be manually
manipulated to implement the requests on the document is both
labor intensive and error-prone.  Linux is built on a
foundation of open standards, which makes it much easier to
parse, manipulate, and transform a specification into an
implementation.  Enterprise Application Integration, ERP, SCM,
CRM, CAD/CAM, CAE, and most other "strategic initiatives" are
much more easily facilitated in the open standards environments
of Linux.  Depending on how you look at it, Windows costs
another 5% of revenue, or Linux saves that 5%, which can be
deployed in projects more focused at generating revenue and
fulfilling orders.

 > Yeah, $150 is nothing to sneeze
 > at, but in the total-cost-of-ownership picture, achoo.

Very true.  Conservative estimates are that the recurring
costs, over a 2-3 year period are roughly 1/4 that of Windows.
  The primary front-load costs of Linux are the issues of
initial training of first-time users, and the costs of altering
IT infrastructure to accomodate Linux users.  This is very
similar to the problems encountered with the first PCs.  The
PCs didn't fit with the SNA LU2/3270 environment very well.  We
eventually switched to Lan Manager, Netware, and with the
Internet, TCP/IP.  Ironically, TCP/IP and the web were
primarily a way to give Windows users the benefits of UNIX
systems through a limited interface, while maintaining
"Macintosh-like" simplicity of a point-and-click interface.

Consider the productivity gains that have happened from 1982
(when IBM released the first PC), to 2001 (Windows 2000, with
many UNIX-like capabilities).  Consider that this is only 1/4
the possible gain available as we move toward Linux and Open
Source based IT technology.

 > I wish we could all run Linux desktops and every
 > server was Linux, but I've become a bit more pragmatic
 > in my old age. Maybe in 5 years we'll be partway there.

Actually, there are somewhere between 50 and 160 million Linux
users world-wide.  Many use "hybrid" systems, either dual-boot,
virtual machines (VMWare), or "Compatibility" (Win4Lin and
WINE) to meet the needs of the Windows oriented management
while maintaining the advantages of Linux.

The question isn't whether, but when, and how.  In 1993, the
"smart money" was on Prodigy, Compuserve, and single-services
dial-ups, and LANs were based on Netware or NetBIOS.  In 1993,
those who suggested that TCP/IP, and "The Internet" would be
hotter than Television or Radio were scoffed as insane.  Many
of those scoffers are now bankrupt.  Many retailers who ignored
the web no longer exist.  Woolworths, Caldors, and possibly
K-Mart have hit the limit because they couldn't see how much of
a difference the Internet would make in our culture.

 > As an additional illustration, in a hosted environment you
 > would think that a Linux platform from IGS would be cheaper
 > than an AIX one. On the contrary, you have to pay for IGS
 > to port their internal support tools to Linux which drives
 >  up the cost tremendously.

But these are one-time costs.  More importantly, there are many
companies such as Sun who have attempted to simplify the
procedure of porting from Linux to Sun.  Perhaps instead of
porting AIX utilities (using proprietary compilers, some dating
back to C-Set2) to Linux, the focus should be on getting these
things working on Linux and porting them back to AIX (using
Open Source compilers such as GCC).

Once those initial costs are incurred, the recurring costs are
actually quite low, and the potential for opportunity is very
high.  We've sold more mainframes using Linux as the foot in
the door, than we sold using MVS.  IBM has a special advantage
in that they can take a Linux package, and scale it from a PDA
to a cluster of Z-900 mainframes.  More realisticly,
developers, such as IGS consultants can do the development of
very powerful and complex systems on R-Series and T-Series
Laptops, in conference rooms, in their hotels, on the
airplanes, in their homes, or in mobility centers, and very
quickly scale them into production environments which can be
scaled from X-Series rack-mount servers less than 2 inches
thick, to Z-Series supermainframes, and even run them in
distributed clusters (beowulf).

When you add technology like DB2, Websphere EE, and MQSeries,
the whole thing starts to sizzle into capabilities we haven't
even begun to consider.  We may not be able to transport the
atoms, but we can mobilize the entire supply chain so that you
can have your equipment, possibly within less than 8 hours, but
easily within 24 hours.  Not only that, but what you buy would
be costomized to your needs.  Whether the product being shipped
was clothing, automobiles, entertainment, or furniture, you
could know before you bought it that you would like it, you
could have it made to order and delivered within a few days,
and you could do it at prices far below those who use the
traditional "off the shelf" approach.

 > The ultimate way to drive Linux on the desktop is to create a
 >  killer application or environment which has great business
 >  value and is too expensive to port to Windows. When there
 >  is a business driver, dollars will flow. So the big question
 >  is, what is the killer app?

 > Answer that question, and I'll be one of the first people to
 >  start developing it.

Rather than thinking in terms of the "Killer App" as a specific
product, consider it's characteristics.

The PC became popular because it could replace the typewriter,
adding machine, and the doodle pad, but what made it really
valuable was that it replaced the ERASER.  If you typed at 60
words per minute, you could type an 80 column, 60 line page in
about 16 minutes, but if you spend 20 minutes fixing typos, or
40 minutes retyping and correcting, you only produced at the
rate of 24 words per minute.  Visual text editors, and WYSIWYG
editors gave you the ability to correct a problem with a single
  keystroke, toking as little as 1/10th of a second.  Even if
you were really sloppy, you could spend less than 30 seconds
correcting that page, as you went along.

Windows, and it's predecessors became popular because you could
combine your drawings, text, spreadsheets, and charts
(generated from the spreadsheets) to create a "publication
quality document", the process became known as "desktop
publising".  You could create your stories, lay them out, add
effects, and then print it on a laser printer, run the master
to the copy machine, and fill up everybody's inbox.

The Internet became popular - and functioned as the "killer app
for UNIX" (the majority of all "Active" Web servers are still
UNIX or Linux) because the mechanics of publishing disappeared.
  You could create e-mail, and a distribution list, and
everybody got a copy instantly, wherever they were.  If they
weren't in the office yet, they would get it when they
connected.  The Web Browser made it possible for publishers to
create a general layout, while the end user decided how best to
configure the final layout based on his needs.  Some users
would dedicate an entire 1600x1200 screen to their web browser,
others would do their browsing in a 640x480 window.  This was
possible because HTML gave them this ability.  More important
than just being able let users access this information, the Web
Browser, when coupled with the search engine, allowed users to
quickly locate information that was relevant and meaningful.
The Google search machine, now consisting of over 1,000 Linux
boxes, provides instant access search over 2 billion pages,
usually in less than 10 seconds, and get 10-20 of those pages
which are most relevant at that particular moment.  And you had
the usenet newsgroups - also known as messaging, which allow
people with common interests to share ideas, solve problems,
and help each other in their persuit of everything from
solutions to world hunger to fixing the latest hardware.

The "instant messaging", such as Sametime, is more of a taste
of what is possible with Linux.  Windows is very user-centric,
it is primarily driven by actions from the user, you can look
at real-time quotes, but you have to poll for them.  Some
investors have investment tools that provide real-time
displays, but the overhead is such that more than a few such
displays can bring the system to a halt.  Many Windows system
become very crash-prone under this type of load.  Linux systems
may slow down a bit, but you can keep on running.

The other key factor in the "Killer App" scenario is the
modular nature of Linux.  Microsoft is hoping that everyone
will spend $billions to convert everything to their version of
XML so that they can convert everything to their proprietary
formats, but nothing should go the other way.  This would mean
that you could send a grid of numbers, and have them displayed
as a spreadsheet, or a graph, if you have the right XSL and
parser to shove the stuff into your screen.

Linux and UNIX consider this an old trick.  With utilities like
  PERL, Awk, Sed, grep, and even Lex/Yacc, they have been able
to quickly parse complex reports and transform it from one
presentation to another, often very quickly and very easily.
When you are only working on a single presentation, you don't
get terribly worried about scripting or automating everything.
  When you are trying to manage 3,000 documents, things get
more complex.  With a modern Linux system on a 1Ghz pentium,
and a 100 Mb/s full-duplex ethernet, you can fetch the 3000
documents, say 2kbytes each, in about 1 second.  But to parse
the content, compile a report, and format it for delivery,
using Microsoft Office tools, could take hundreds of
staff-hours.  You could distribute the load.  You could have a
set of managers, each managing 5-7 people, who give him weekly
reports - each taking 1 hour to compose, format, beautify, and
match the visual look of the Moore Business form once used back
when computers were fed by key-punched cards punched by
near-sighted operators who had to decipher glyphs on a
handwritten form, he could compile those into his weekly report
- figure 1 hour/subordinate, hand them his supervisor, who
manages 5-6 managers, who hand them to their managers, who
manage 4-6 subordinates, let's just assume an average of 6
people per manager, and 6 levels.  The CEO has 6 presidents,
who manage 36 VPs, who manage 216 directors, who manage 1296
managers, who manage 7776 line workers, for a total of 9330
employees, working an average of 2 hours/week "making pretty
reports and cutting out the "noise".  That's over 18,000 hours,
at an average of perhaps $40/hour, or $720,000 per week, spent
on manually keypunching what you did, and what you will do.

Suppose instead that all of this were something that could be
checked off on a web-page.  You could have each sales person
click a counter for each call made, for each sale closed, and
enter the amount of the sale - already synchronized with the
person he is calling.  This gets fed real-time to the order
processing, accounting, and management functions.  If there is
a problem fulfilling the order, actions can be taken within
minutes of receiving the order, not three weeks after the
shipment was due to arrive.  The CEO can see the accounts, with
a real-time chart.  He can see the problems, the same way the
network manager on an OpenView console (Linux or UNIX
workstation/console) can identify a faulty circuit, long before
anyone even notices there is a problem.

This means that management can be more focused on decision
making, causing action and results, and implementing the
solutions, and less time focused on "making bad news look pretty".

Linux is the invention of the "Slacker generation", these are
the folks who had to create their own businesses from the
ground up, worked at cyber-cafes, and did business by
cell-phone, laptop, and internet, back when most of the leading
edge baby-boomers were still printing mountains of hard-copy
and dumping it in everybody's mailbox.

These were the folks who weren't able to count on high salary
jobs that lasted for 20 years, homes that always appreciated,
and 401K plans that automitically appreciated.  These were the
folks who often worked as contractors, started their own
businesses, traded online, sell short, and day-trade.  These
are the folks who help create dot.coms when they were hot, and
were taking "the Web" into the corporations when their baby
boomers were coming in with their high-dollar budgets turning
out the "dot.bombs".

These are the people you want as customers, as employees, and
as partners in projects.  And they do like Linux.

Rex