Post by r***@usa.netPost by billwgPost by John KloostermanWell - the problem here is the therm "market-share". It is perfectly
posible to see how many pieces of Windows OS are beeing sold, with Linux
however it is a different story. Most versions (distributions) of Linux
are free for download, so the users dont have to buy or register anyware.
That makes it almost impossible to figure out how many people are using
Linux. What makes it even more complicated is the fact that there are many
pirated versions of Windows, ?nd a lot of people using Windows and Linux
in a mixed environment or as a double-boot system.
Market share statistics can vary radically depending on what you are
counting.
Count how many machines are "sold" with Linux preinstalled, and the
number is almost microscopic (since most Linux servers are configured
by the corporate administrator). Ironically, many of the servers now
running Linux were originally "Sold" as Windows machines.
You can count number of servers, which puts Linux at about 30% and
Windows at about 40%, with UNIX at about 30%, depending on the year
sampled.
You can also count revenue, but again, this will skew the numbers
toward Windows - especially if you make the distinction between License
fees and Support fees and do not include support fees in the count.
Again, because Linux usually gets the "leftovers" or "hand-me-downs",
Linux will be consistently be undercounted here too.
A very interesting approach is to count "function points" or
"processing equivalents". A single Linux server typically does the
work of 3-4 NT servers. A Solaris StarFire or HP Superdome can do the
work of 5,000 NT servers. A Z-Series e-server running ZVM and Linux
can do the work of 50,000 NT servers.
Post by billwgPost by John KloostermanSo -i think its impossible to get accurate figures. You can only use
statistic methodes, but then it depends on the willingness (and
acuressness) of the "statistical group" involved. It leaves also the door
wide open to "bias" the outcomings (as has happend in the past and w?ll
happen in the future).
It's possible to get accurate figures, just very expensive. There are
services who do detailed studies, and gather the numbers a number of
different ways. Often, "abridged" reports are published about 6-12
months later for use by advertisers.
Post by billwgThe term "market share" only relates to the share of some identified market
captured by some product provider to that market. It really only pertains
to whether or not your product has the largest share of that market because
your position in the market somewhat dictates your strategy and what the
future might hold for you based on market characteristics that have been
determined to be predictable from the past. The most important thing that
you can do is to determine just what your "market" consists of.
This is why it's so important to know the exact methodology used in the
survey.
It's also necessary to have the raw numbers as well as the various
percentages. It's also important to have the breakdown of anything in
the "other" classification.
One of the great classic surveys is the "ip/browser" counter survey. A
company did a survey of something like 500 sites, all running Windows
based servers, and doing "browser counts". Further disclosure of the
details showed that this company was only counting Linux on IP
addresses that showed ONLY Linux. Furthermore, it was counting DHCP
pools indescribinately as well. In this survey, Linux got less than
1%, but this was ONLY browsers that had "Linux" in the browser
signature. A bit of research into the 3% marked "other" showed that
nearly all of the "other" turned out to be Linux because Netscape and
others were not specifically placing "Linux" into the default
signature.
Post by billwgThe market for "Intel compatible desktop client operating system software",
which is the market area that Microsoft was found to have monopoly power
within, was determined by the Jackson court to specifically exclude Apple
computers and their software as well as Linux computers.
I think that even at the time that the trial started, Apple had less
than 5% of the established market share. Microsoft claimed during
their defense testimony that Linux had 14% of the market. It's quite
possible that Microsoft was counting cookies against signatures and
tracking all permutations. This would be about the only reliable type
of count. Even then, you might get dual-boot machines counted as both
Windows and Linux (since there would be two different cookies), or
VMWare images.
Post by billwgApple was excluded
then because it was not Intel compatible and Linux was excluded because it
was a server product.
During the trial, Microsoft did try to give Linux a much larger market
share. I found it significant that no one actually challenged them. I
might have been grounds for challenging the findings of fact as well as
the findings of Law.
They just dropped a big number and everybody let it slip.
Post by billwgGiven that definition, Microsoft has 100% of the
market or very close to it, since today there clearly is some linux aimed at
desktops and some money has changed hands on that basis.
One of my favorite numbers is the number of licenses Microsoft claims
to have shipped/sold, and the number of PCs the OEMs claim they have
shipped. Microsoft has about 120% of the market by that count. I know
from the period of time I was setting next to the cubicle used by the
Microsoft rep to make calls to a number of customers and decision
makers that the sales rep was offering substantial discounts to
corporate customers when they purchased more licenses than they
actually needed.
Of course, since Microsoft often oversells the OEM and then oversells
Windows licenses to Corporate customers for the same machines - it's
even conceivable that Microsoft has over 200% of the market. And since
Microsoft donates tax deductable Windows licenses to 501C organizations
when the machines above are donated to charities, Microsoft might even
have something like 250% of the market. If you think about it, it's
really quite a feat of marketing.
Post by billwgThe money is what makes the market. Markets are measured in terms of the
money spent to buy products that compete in the same market and market
growth or decline is measured in terms of money increases or decreases over
time. It has nothing to do with how many people use the products or how
often they use them. It only depends on how much money they spend annually
for the products.
This certainly gives Microsoft the advantage. If you figure Microsoft
licenses and cost about 3 times the price of Linux licenses, and that
Linux hardware is mostly hand-me-down hardware from Windows, Windows is
certainly going to have the biggest dollar sign percentage.
Post by billwgIf a customer purchases a linux based PC to replace a Windows based PC, that
is a loss of share for Microsoft.
The more interesting case is when a customer purchases a new Windows PC
to replace an older Windows PC, then the old PC is converted to Linux.
The practice of purchasing a new machine for Linux seems to have
increased since the availability of AMD-64 bit chip.
Post by billwgIf the customer switches from Sun servers
and terminals to some distributed linux based PC network, that is a growth
of the desktop market as well as a share loss by Microsoft, but it is also
an opportunity for Microsoft to compete for the business.
This one is more tricky. If someone were to upgrade from a group of
10-12 Sun servers with 2 1 billion instruction per second processors,
to a StarFire with 64 processors each running 4 billion
instructions/second or roughly 128 times more power, they might not
need so many servers. By server count, Sun is losing market share, by
cost, the supporting hardware of the separate machines would make the
new server cost less - again, Sun loses market share. Of course, in
reality, the Sun box is actually doing the work of more machines.
Post by billwgPost by John KloostermanSo - i guess there wil be no conclusive answer. I think however that the 1
percent mentionned is a far to low figure. In my neigbourhood i have seen
an strong increase of Linux desktop usters in the last half year, so i
think that figure is very dated...
Again, it depends on what you are counting. The range goes from less
than 1% of all the machines sold in the United States by OEMs, to as
high as 35% of the server market based on number of servers. If you
really want a stretch, you can figure Linux is almost 40% based on
"function point" counts - essentially "NT equivalents".
Wow... excellent, and startling observations!