Slashdot Mirror
New Numbers on Linux Market Share Soon
prostoalex writes "New numbers on Linux market share are due this week. As far as global PC market is concerned, Gartner claims 5% of all PCs shipped this year ran Linux OS, although by the time the PCs were actually on the user's desk, only 2% of them run Linux. In the server world IDC estimates that Linux-powered servers comprise 28.3% of all server sales in 2004."
Gammage also stated that until Linux is shown to support the NX (No eXecute) security technology supported in Microsoft Corp's forthcoming Windows XP Service Pack 2, it will be seen as potentially deficient to Windows. However, Red Hat released a patch for the Linux kernel to support NX in June that has the full blessing of Linux creator Linus Torvalds.
Yeah, right. Read 'em and weep.
"Lawyers are for sucks."
- Doug McKenzie
The price of the licensing isn't important, the TCO is. It's about how much you value your time. Linux makes you much more efficient.
Now of course, there are some machines where a linux install is a REAL pain, but most on "desktop" hardware sail right along.
Of course, home builts arn't counted in these numbers either so you've still got new systems running linux that weren't accounted for.
I work for a network installation/support company, and trust me...there are plenty of people that will be running server installs longer than 4 years. We just recently had to drag a customer (kicking and screaming at that) through an upgrade from Novell 5.5! I know for a fact that we have three customers currently running NT 4.0 (one's still running 3.5!) and probably won't upgrade even in Microsoft discontinues support. There are plenty of Windows 2000 servers around here running like tops, and I don't see any trouble with them lasting for a good 3 or 4 more years, barring any catastrophic hardware failure.
If I'm not mistaken, Dell offered business users a choice of some random flavor of Linux or FreeDOS preinstalled on their computers.
The logic was thusly: Licensing agreements force them to ship the computer with SOME OS on it, but a lot of businesses already have a Windows site license. Because the price of the non-windows Dells was slightly lower than effectively purchasing the license a second time, the companies order these computers with one of those two OSes preinstalled, then wipe the drive and install WinXP/2000.
So while there are a small number of users purchasing their computers with Windows pre-installed and migrating to Linux, that number is effectively buried by businesses doing just the opposite.
People who buy from Dell and don't want to pay for Windows. You can't buy the machine with no OS, so you choose Linux (which is cheaper than Windows) and then just install your warezed copy of XP over the top.
How we know is more important than what we know.
Gartner is in the business of selling "reports" and "studies".
Most of the "reports" and "studies" you'll see from Gartner are linked from vendor's websites. Vendors who paid for the report. So the vendors use those "reports" and "studies" as marketing materials.
I've only seen Gartner stuff used to justify a decision that has already been made. And, IMO, that's all they're good for.
I fake my headers to say "IE/Windows". It's amazing how many sites suddenly start working when you do. Don't most other Linux users do this too?
Still, I know nobody who bought a computer with Linux. Altough I know quite a few Linux users.
In India about 5-10% (probably closer to the latter figure by now) PCs are sold with linux pre-installed. Obviously, not all stick with linux. My guess is 2-3%.
Its the same situation in most of Asia. Linux PCs are reportedly selling like hotcakes in Malaysia. In China, it is even more extreme than in India because the number of people actually using linux is negligibly small.
The reason for this is that most home PC users in these countries use pirated software whereas OEMs still have to pay for Windows if they want to install it. The amount of wipe-out-linux-and-install-windows going on in Asia totally dwarfs the number of geeks in the world installing linux on their machines after paying the windows tax.
Slashdotters are living in the 1990s. The new reality is vastly different from what it used to be. The vast majority of linux users are non-geeks. There is no problem at all in getting linux PCs. The number of Linux PCs sold significantly overestimates atual usage.
The reason that linux usage continues to hover around 2% is no longer due to Microsofy bullying, but because Linux is still quite hard for non-geeks to use.
When on Google, my debian system with Konqueror defaults to stating it's Windows XP with Microsoft Internet Explorer 6.0
I can change it, but how many other browsers falsely identify themselves for the sake of extra security against the internet?
The Penguin Producer
Try: http://www.linuxcounter.org/
Which has been around for about 10 years. Not sure why it isn't more popular. Would be nice if Linux distros gave a choice at install to "auto-register" or something. There is also good info there on why estimates on number of Linux systems is difficult to guage.
Also:
"Market Share" means number of units sold vs the total number of units sold in a particular category during some period of time (year quarter whatever). MS and even more so, Apple, have a very good idea of what their market share is. They know almost exactly how many of their units have been sold. Linux has a pretty small market share because everyone can just download it once and install many times. On the other hand comapnies like Red Hat probably have a very accurate count of what their market share is for RHEL.
"Installed Base": Means the number of particular units installed vs the total number of units of all systems installed of all types for a particular instant in time. The installed base is certainly higher for Linux than the market share is. Installed base is probably quite different from market share for all types systems. There are all those PCs, Macs, and Sun boxes that now have Linux installed. There are all those homebuilt systems that now have a pirate copy of Windows or Linux or both installed. There are all those PCs that shipped with Windows (or Linux) that now have Linux (or Windows) installed or both. There are all those systems that can't run the latest version of Windows but are now running Linux in "edge-of-network" roles or just as plain old desktops.
Apple probably has a very good idea what their installed base is for Mac OS X (compared to their market share) because of Softwareupdate.app grabbing the latest updates which, pretty much exclusively, come from Apples servers.
MS may have a good idea of, at least, how many people update their systems via windowsupdate.com (or whatever it is called).
Linux distros may not have a very good idea unless they can somehow get stats from ALL their update mirrors. But this would never include all the locally mirrored updates. I download updates from one to three mirrors once and then push those out to dozens of systems locally.
"Mind Share": Though a popular buzzword has nothing to do with market share or installed base.
While Linux market share, however it is measured, should give a company a solid number to work with when determining the size of their market... it says nothing about their real and potential market. The Linux market share is the lowest possible number to use for estimating the size of the potential Linux market for a company that plans to release a product for Linux systems. This is a very important thing for a company to know when planning a product release.
So if you want to see more products for Linux... go get counted.
-DU-...etc...
"Don't sweat the technique."
Your statement that most "geeks hit the same sites as non-geeks" is non-provable and, without a valid premise, your metric is also invalid.
Being non-provable is sufficient, but to not just leave it to the trivial counterproof: you're metric does not take into account that most browsing is done during the day which for many is at work which means that your are getting more of a corporate browser and OS measure than anything else.
Which in my case would count the WinXP system I use at work for my primary desktop, but miss the two OS X desktops and the RedHat desktop, not to mention the two SuSE desktops and OS X laptop at home (and, no, that's not trying to count in the two additional SuSE servers as desktops).
I stand by my original claim: there are no metrics.
Thoromyr