The author based his article on distrowatch statistics.
Of course, not everyone visits distrowatch (I do about once every 3 yrs or so)
I can’t even guess the profile of the person who would even visit distrowatch (I’ve only ever for curiosity about the strange and unusual).
But, that got me to thinking, is there even a recognized source somewhere that might measure usage of various Linux distros?
I don’t know you can even use source install download metrics reliably since with no restrictive limitations copies can be made and used at will.
If no one is actually tracking distros, the closest I can guess at measuring overall distro usage is tracking versions of web browsers used by Linux in general.
Not using Distrowatch but I check from time to time Wikimedia Traffice Analysis Report: http://stats.wikimedia.org/wikimedia/squids/SquidReportOperatingSystems.htm and from the collected numbers I would say that most Linux users are working with Debian or Debian based ones (Ubuntu, …) by a large margin. That numbers are true for end users but however in bigger environments RedHat and Debian are dominating - which is not surprising as with a clone of RedHat (you know which clone I mean) it’s very easy and good for RedHat to keep new users or for whatever reason which to do not go with the official red one but with RedHat technology. I think it was clever from Redhat that they never put more rocks in the way to make such a step impossible and they get a lot of testing, feedback and new developers for “free”. That’s one of the things which I personally miss from SuSE - do offer SLES/SLED the same way - we got Evergreen but that’s not the same. Sure you can generate somehow an identical image with SuSE Studio but as I have tested it it’s very hard to keep a generated image up to date this way. However I would jump for sure to such a project in case openSUSE evergreen get turned to a free clone of SLES (I’m only dreaming as I think this will never happen). Debian has I think the largest community and it’s very easy to use it as template for in-house projects.
Having read a couple posts on this, I think the biggest problem is maintaining the free clone and the organization required to even initially set it up. It would be a large project and I dont think there is enough demand for it that it could survive for long.
I’ve heard criticism that Distro Watch is less “scientific” and more of a “popularity contest” from a small set of people.
I go to Distro Watch every so often to check on a distribution I read about elsewhere but that is only because I want to find what version of something they are using or their official project website. That should hardly count as my preferred distribution.
Some distributions keep track of the number of unique IPs that download their installation media, but I am not sure if they track distribution version update and it doesn’t take into account people, like me, that sometimes download the DVD once and use that to install the server, desktop and laptop.
Even tracking location of updates isn’t fully accurate because not everybody updates it all the time.
So there isn’t really any ONE accurate place to track and determine “market share” of different distributions. Maybe Market Share of Enterprises can be tracked in the usual fashion but outside of that it is a little tougher (ok, Super computers may be easy to track too… and Linux has something like 94.6% “market share”)
One problem is that a money-making organization have different emphasis (long-term stability and security) than hobbyist and many desktop users (latest software, new gadgets, etc.)
Red Hat handles this with the enterprise version (Red Hat, CentOS, ********** Linux) and the desktop/development version (Fedora). Ubuntu handles this with LTS releases focused for enterprises and in-between versions for development and bleeding edge.
I like the idea of there being one “SUSE”, but seeing the wrangling Ubuntu and Red Hat are going through, I don’t think it is reasonable at this point. It would take some MAJOR shaking up of organizations.
I think that people generally understand that Ubuntu is by far the most popular Desktop, and I don’t have a problem with that.
I generally advise people to choose their distro based on features matched to needs, and of course openSUSE has done a good job most of the time satisfying my needs.
I don’t know if this is an accurate way to estimate the Wikimedia’s statistics, but I found it interesting it breaks out Android versions and I know a fairly reliable source to compare for that information… The following is updated monthly for the benefit of Android developers who have to know how to allocate their build efforts to support as many Android versions as they want
Comparing the market stats from the Android Developer site with the Wikimedia stats, they mostly agree only a general ordering of popularity, the degree any version is more popular can be very different.
So, as might be expected it seems that despite reading stats from one of the Web’s most broadly visited websites (Wikipedia), there seems to be limitations. In the end, the data is only exactly what it is, and cannot be extrapolated to the rest of the Internet.