The State of Linux Distributions: an analysis of what the Distrowatch rankings tell us
I have always been interested in Linux distributions and how they evolve, not just technically, but also in terms of their popularity and their -wanted or unwanted- position among the Linux users and on the broader market.
In this regard, the Distrowatch web site is a fairly effective tool at tracking linux distributions no matter how small or short-lived they are. I think the site's online tool is actually the best one so far, and it has, in my opinion, produced the best stats concerning Linux distributions, unless you consider the NetCraft surveys but these are confined to the web servers' world. Distrowatch takes all of them, servers, desktop, set top boxes, and now has come to include OpenSolaris and BSD Unixes. It spans over 100 distributions, although there are more (but then these are dead or fairly underground distros).
What I would like to do here is to give the interested readers a rather personal outlook on how the Linux distributions have evolved in terms of popularity, which are hot ones, why, and what I noticed behind the daily, weekly and monthly rankings of Distrowatch. I have been able to uncover some interesting patterns, and ones that are not often talked about.
First, I should say that the Distrowatch "Linux quotation engine" as I call it, has evolved recently and is now able to provide us with much more granular insights and data. But as long as the stats have not been consolidated for one year (they do this typically at the end of each year), you should always keep in mind that the trends (up, down or idle) always represent the ratio between the various results obtained on a given period of time. For example, let's say distribution XYZ starts the year with lots of momentum but goes south at the end of the same year. It may be that it will remain at the same place considering the number of hits (the site has very accurate rules for measuring those) but you will see the trend go down if this has proven to be the ratio in terms of hits between its first momentum and its actual status. The time range that I shall consider here is the year 2005 - may 2006, but some background on 2004 will also be taken into account. One last comment however: Distrowatch's seven last days rankings are difficult to understand if you do not take into account that it tends to reflect a more immediate image of the visits to the site and the newest trends that are generated by special factors like a new release of a distribution.
That said, let's go straight into the heart of the matter. Unless you have been living under a rock, you have certainly witnessed the outstanding rise of the south african Ubuntu, and this ever since 2004. Of course, this shows well in the rankings since Ubuntu is number one and has risen very quickly from the depths of the quotes to the front row. Now even Ubuntu's own derivatives, Xubuntu, Edubuntu, and Kubuntu are listed and fare pretty well. The mother distribution, if it is still the case, Debian, ranges between the 6th and the 9th positions, but remains on the overall idle, neither losing nor winning shares. Suse and Fedora come close, followed by Mandriva and the other well-known Debian derivatives (Knoppix, Mepis, Damn Small,etc).
Well-established distributions and distributions that have written the history of Linux (Slackware, Gentoo, etc.) typically ranges just below Debian.
What I found by looking at these rankings for several years now is rather interesting. To give a brief overview, there are distributions who do not move or not so much (say from one place up or down in one year) and they typically range from the 5th to the 10th positions. You will notice here that I am thinking about Debian, Slackware, Gentoo and the likes. Below the 10th or 15th ranks, the story does not become less interesting, but certainly rather chaotic: there is simply no assured ranking in this area. Distros come and go, sometimes up, sometimes down. The only notable exception is Red Hat (not Fedora), but this is is explained by the actual Red Hat business model where only the official Red Hat customers who would visit that site are listed on Red Hat, hence its rather disappointing 24th position, but only disappointing at first sight. I typically consider that the movements and the trends of the distributions start to become interesting when they reach the 30 zone. Below the story would be rather anecdotal, no pun intended to the users and the authors of those distributions.
Now you may wonder what is going on from the 5th to the first place. There things become hotter and hotter the closer you get from the top, and although it is possible to depict some clear patterns, few distributions are able to join that fray, although it is possible for some of them to get out of it. As a sad example, I have clearly seen Mandriva fall from the first place it had for years, to be chased out from the top and bearing constantly down for more than three months by Ubuntu, and Suse and Fedora as well. Actually, this could mean a lot more than any other financial stats one could find about Mandriva. Even under their bankruptcy period, they were still the number one distribution so that at that time you may have understood their situation to be a pure problem of business model definition. Today, it is worse. By reaching the 6th place this month and still bearing down, Mandriva got out of the "magic 5" and gets closer to its derivative, PC Linux OS that fares very well at the 8th position, separated from Mandriva by Debian (but things should not last for PC Linux OS as well).
What I see in the Mandriva story today is that they not only did lose market share, but that they did not manage to come back with innovative ideas and technologies, and that reflects clearly on their rankings. Ubuntu has replaced Mandriva, and you can check that yourself by enquiring on this around you.
A not so big surprise may be the popularity of Suse and Fedora. This has pretty much stood true from Fedora ever since its inception, as it rised up from the bottom to the top in a constant and gentle way. The way I see it evolve may mean that the third place may be Fedora's constant position in the Linux world, and that's a fairly nice number, at that. Suse is a different story, though. Suse always had good rankings but it is now maintaining itself to the second place and from what I see, its latest releases and the opening of a stronger community wanted by Novell might actually make things happen. Not only Suse gets the number two, it also eats Ubuntu's own market shares, and this would explain a rather constant, yet light trend down visible on Ubuntu's ranking (one could object this erosion comes from Ubuntu's own derivatives but I am not so sure about this).
But the really big surprise to me comes from the nebulous 7th to 20th region. First, on the Debian side of things, a rather unexpected trend I see is that in terms of rough numbers, Debian itself as a distribution did not get that much hurt by Ubuntu. While it is true that it did not take it to the "magic 5" Debian either loses a a few market shares, or eats some of them itself. Debian always fares between the 6th and the 9th position, and do not move away from it. Derivatives such as Knoppix or Mepis are clearly hurt although they still range around the 5th rank, with the notable exception of Kanotix, who, despite a general awareness in the community, doesn't seem to find a broad audience.
Second, and that comes to me as a real and unexpected pattern, the success of the Slack family is rather unwitnessed to many people, but will end up being acknowledged. Slack is back thanks to its official or unofficial derivatives, SLAX, Zenwalk and others. While Slackware itself remains a pretty unchanged value between the 11th and the 15th position, its live CD, SLAX, has managed to go up to the 10th position, showing a rather broad community of users. Vector also goes constantly up and Zenwalk is gaining some traction, albeit well under the 15th position. The Slack momentum is rather surprizing, and could be explained partly by the recent birth of end-users focused derivatives, but it may not explain it entirely. Maybe Slackware's own quality and community responsiveness could? This is an open question.
Last but not least, I have three other distributions to watch closely, as they say on stock markets' live reports: Gentoo, Arch Linux and Frugalware. Frugalware's rather disappointing performance is explained by its youth and the fact that although it has gained momentum its community is still to small to fare up well on Distrowatch. Arch Linux goes up and pretty well from the bottom up to around the 24th position. Gentoo, however, goes down despite several initiatives and the opening of new derivative distributions.
All in all, I believe that this analysis gives a detailed report on some key distributions under an unusual angle. I welcome questions and critics as there may be finer distros' analysts that me out there. In any case, keep your eyes stuck on the rankings !