TuxRadar by the numbers 2.0: the rise of Karmic


Many moons ago we posted some stats about TuxRadar, explaining the breakdown of our visitors, which pages were proving popular and more. That article proved popular enough with you folks that we think it's time to rinse and repeat to see what we find this time. But - thanks to sifting through 32GB of logs - we have some even more interesting numbers for you this time, showing that Ubuntu 9.10 adoption is outpacing that for both Windows 7 and Snow Leopard...

Let's start from the top: our most popular articles in the last few months have been:

  1. Benchmarked: Ubuntu vs Vista vs Windows 7
  2. OpenSolaris vs Linux
  3. Linux vs Windows 7
  4. Vista, Windows 7, Ubuntu 9.04 and Ubuntu 9.10 boot speed comparison
  5. Ubuntu 9.04 32-bit vs 64-bit benchmarks
  6. Ubuntu Netbook Remix 9.04 hands-on

The fact that Windows 7 appears in three of the top four articles on our site should come as no surprise - we already posted about the surge that Windows 7 is bringing to Linux, and there's your proof right there. Of course, it's also clear that Ubuntu also appears in four of the top six; more on that soon.

As with last time, the long tail means that those top six articles make up only a small part - about 15% - of our total page views. To put that into perspective, our free online PHP e-book has three times more page views than those top articles combined. Articles such as Make Linux faster and lighter, How to choose the right Linux distro for you and How to set up your own Linux media server make up the remainder of the page views.

People are finding these articles in a variety of ways, but recently Digg hasn't been sending much traffic our way. From what we can tell, general Digg visitor numbers seem to be down, which makes it harder for Linux topics to have enough up-votes to become popular. Still, Google has risen to the challenge admirably, and now sends us 37% of visitors. We're always glad to see visitors from Linux Today, OS News and Tux Machines - these tend to be quite technically advanced readers (our favourite!) and usually surf through a few pages before leaving.

Much as we'd rather not have to say it, Windows users continue to be the majority around here, and in fact they have grown slightly. Here's the breakdown, with our previous figures in brackets:

  1. Windows: 55.47% (55.11%)
  2. Linux: 35.95% (34.47%)
  3. Macintosh: 7.29% (9.19%)
  4. iPhone: 0.38% (0.45%)
  5. Unknown: 0.36% (0.35%)
  6. SunOS: 0.17% (0.07%)
  7. iPod: 0.11% (0.14%)
  8. Android: 0.08% (0.05%)
  9. FreeBSD: 0.08% (0.08%)
  10. SymbianOS: 0.05% (0.03%)
  11. OpenBSD: 0.01% (0.01%)

As with last time, even if our Linux users increased by 50%, they still wouldn't match the Windows number.

Now, here's where the really interesting stuff can kick off. As you can imagine, the users we're really interested in are those Linux users, so we wanted to see just what kind of stats we could pull out of our logs by using a few standard commands. To start with, here's what an average page request looks like in our logs:

xxx.xxx.xxx.xxx - - [08/Nov/2009:06:47:38 +0000]
"GET /content/benchmarked-ubuntu-vs-vista-vs-windows-7 HTTP/1.1" 200 93519
&q=ubuntu+vs+windows+7&sa=Search" "Mozilla/5.0 (X11; U; Linux i686; en-US; rv:
Gecko/20091020 Ubuntu/9.10 (karmic) Firefox/3.5.3"

There are a few interesting bits in there:

  1. We've anonymised the IP address to xxx.xxx.xxx.xxx.
  2. This person visited our Benchmarked: Ubuntu vs Vista vs Windows 7 article.
  3. They came from the Google search "ubuntu vs windows 7"
  4. More specifically, they were on the Ubuntu 9.10 Google start page - that's the partner-pub bit (although it irks us that the page refers to it as Ubuntu 9.1).
  5. They were using Firefo 3.5.3 on Ubuntu 9.10.
  6. They were running a 32-bit version of Ubuntu - that's the i686 part.

So, when you dig around, you can actually tell a fair amount about visitors - as long as they identify themselves clearly. Not every distro does identify itself so nicely as Ubuntu, but it turns out that the following do:

  • Ubuntu 10.04
  • Ubuntu 9.10
  • Ubuntu 9.04
  • Ubuntu 8.10
  • Ubuntu 8.04
  • Fedora 12
  • Fedora 11
  • Fedora 10
  • Mandriva 2010
  • Mandriva 2009.1
  • Mandriva 2009.0
  • PCLinuxOS 2010
  • PCLinuxOS 2009

There were a couple of other distributions that we thought we could probably guess at, but we're not really interested in guesswork: the distros above identify themselves conclusively.

So, with a list of distros we can identify conclusively, plus 32GB of log files and a lot of time on our hands, we decided to dig through and see what we could find. Let's take a look...

(NB: our site numbers fluctuate, so we're going to be talking in percentages; we used logs from our server up until November 8th)

Ubuntu continues to dominate

Let's start with a simple chart: if we add up all Linux users - identifiable and unidentifiable - how many are using Ubuntu?

How many visitors are using Ubuntu?

The answer is an average of 54.8%, which is pretty much the same as we reported last time. Shortly after Ubuntu 9.04 was released, that number spiked up to just over 65%, but it declined pretty much from then onwards until it reached its all-time low of 47.32% just over a month before Ubuntu 9.10 came out. Karmic doesn't seem to have had such a major impact as Jaunty, which perhaps isn't surprising given some of the negative press it has received.

NB: 54.8% is a minimum figure who use Ubuntu - it doesn't include spin-off Ubuntus such as Kubuntu or Xubuntu, nor any Ubuntu user who changed their user agent or installed their own web browser.

Of course, if Distrowatch has taught us anything it's that distro spikes occur all the time; a more interesting graph for us is this one: which versions of Ubuntu are people using?

Which Ubuntu versions are most popular?

Now that is an action-packed graph. In fact, we could probably have removed the legend and you could still have figured out which line was which. There are a few interesting things to note here:

  • Neither Ubuntu 8.10 or 9.04 did more than graze gently against the 80% mark - clearly not everyone is eager to take advantage of the six-monthly update cycle.
  • A large group of people start testing the next Ubuntu release as soon as the beta is announced, rising further when the release candidate is announced.
  • With six months still to go until the next LTS release, Hardy (the previous LTS) release has retained under 5% of the desktop market share. The drop just as Jaunty is released is quite telling - until that point it had a healthy 20% market share, but that promptly halved after Jaunty came out, presumably as people started to prize new features about stability.
  • You can barely see it, Lucid Lynx - Ubuntu 10.04 - is already starting to gather some users - a mere 0.08% of Ubuntu users so far, but it only started appear in our log files from November 1st onwards so there's a long time yet.

But beyond all that, what we're interested in is the lightning-fast adoption of distros. You see, ten days ago the top bod of TechRadar (a popular mainstream geek news site) announced that "Windows 7 is already 13% of visits to TechRadar - Vista took 2 years to get to 22%." Of course, it didn't take long for our friends over at MacFormat magazine to reply, saying "And Snow Leopard accounts for 70% of MacFormat visitors. Keen upgraders, those Mac people.". To be fair, that's not really a good comparison: Windows 7 had come out one week after the Windows 7 stats figure, whereas Snow Leopard had been out for over two months by the time MacFormat looked at its numbers.

So, if Windows 7 took 7 days to get to 13% and Snow Leopard took 70 days to get to 70%, how is Ubuntu 9.10 doing? Well, here's your answer: after just 10 days, Karmic already accounts for 64.4% of all Ubuntu visits to our site. That's about seven times the adoption rate of Snow Leopard, and leaves Windows 7 in its dust. So: neener!

Fedora: quietly innovating

Such a huge amount of innovation in the Linux world is done by the Fedora team, and yet we still struggle to understand why its market share remains so low. Still, it is number two amongst our visitors, but again we should remind you that we're measuring only the distros we can track. Let's take a look at the breakdown of the three most recent Fedoras, including the just-released Fedora 12:

Which Fedora versions are most popular?

Forget the numbers for a moment, and instead look at the shape - it's quite different to the smooth bell curve graph of Ubuntu users, and indeed Fedora 11 seems to have suffered an early drop in users just before Ubuntu 9.04 and Mandriva 2009.1 were released. Our Fedora usage peaked at around 13% of visitors, but that has since dropped back down again, and in the current distro climate - Ubuntu 9.10, OpenSUSE 11.1 and Mandriva 2010 all jostling for attention - Fedora 12 is going to have to work very hard to pull back some users.

As with Ubuntu, Fedora users seem eager to jump onboard the latest and greatest distro as soon as it's released: Fedora 11 quintupled its market share in a week, then almost doubled it again in the following two months.

The long tail of Linuxes

Originally we had hoped to show one big chart of all the identifiable distro visitors to our site, but that got nixed early on simply because the size of Ubuntu basically flat-lined almost every thing else. And so we're going to dedicate a chart just to Mandriva and PCLinuxOS. Here goes:

How many people are using Mandriva and PCLinuxOS?

Again it's easy to see the spikes and dives that occur when new distro releases occur even in this shallow end of the pool, but there's more to look at here:

  • Mandriva used to be the top dog in the distro world, and now owns about 2.5% of our identifiable visitors. That's about 1.7% of our total Linux visitors, identifiable and otherwise. Is this going to be Ubuntu's fate one day? Certainly Linux users don't seem to have too much loyalty to a distro when shiny features abound elsewhere.
  • PCLinuxOS is a pretty small distro, but it's also a really small team. What's more, PCLinuxOS hasn't even had an official 2010 release, and yet its users have already installed the latest packages to effectively bring themselves to where the 2010 release would be - now that shows where the really eager Linux users are.

And finally: 64-bit Linux still sucks

Linux was one of the first operating systems out of the door with full support for 64-bit x86, more than six years ago. How are we doing? Here's the chart:

How many people are using 64-bit Linux?

As you can see, on average less than 25% of our visitors use 64-bit distros. Yes, the numbers are moving upwards ever so gently, but the rate of change is glacial at best. This is a source of worry for us in the long term, because Microsoft is pushing towards 64-bit with Windows 7 by allowing only 64-bit-capable PCs to be marked as Windows 7 ready, and most Macs running Snow Leopard are running most of their apps in 64-bit mode. Linux, it seems, is lagging behind.

What does it all mean?

We can draw a few conclusions from these numbers:

  • Generally, Linux users jump to the latest distro as soon as - and often before - it's available. Yes, Ubuntu 8.04 is supported on the desktop for another 18 months, but we don't think anyone will be using it by that time.
  • Adoption of Ubuntu 9.10 has been incredibly fast, and will probably pass 80% by the end of the year. This is substantially faster than both Windows 7 and Snow Leopard.
  • Uptake of 64-bit distros continues to be disappointingly slow. We're really hoping that Ubuntu 10.04 will make 64-bit the standard choice when downloading, because that's the kind of thing we'd need to see in order for Linux users to move to 64-bit in the near future.
  • More Linux distros should identify themselves clearly. We'd love to see OpenSUSE on these graphs, and we'd love to see Red Hat Enterprise Linux too.

We can also use this vast quantity of numbers to make a (pretty-safe, we feel) prediction: Ubuntu 10.04 will almost certainly be the first Ubuntu version since 4.10 to pass 90% market share among Ubuntu users, as the last remaining desktop LTS users will have to jump ship along with everyone making the upgrade from Karmic.

You should follow us on Identi.ca or Twitter

Your comments

55% of all visitors on a

55% of all visitors on a Linux site are visiting from a Windows system? That's quite peculiar...


See the comments on the previous story - a lot of folks took the time to explain why they (have to) use Windows.

Not that peculiar when you

Not that peculiar when you take into account the fact that the majority of these are probably accessing the site from work, where Windows is king.


When I'm at work I use Windows. I hate it, but looking at a Linux site keeps me sane.

Running Ubuntu 9.10 now, but I'm looking at Fedora 12 with a keen eye.

@Nobody Important

3-second review of Fedora 12: OH MY GOD IT'S FULL OF AWESOME.

"We're really hoping that

"We're really hoping that Ubuntu 10.04 will make 64-bit the standard choice when downloading, because that's the kind of thing we'd need to see in order for Linux users to move to 32-bit in the near future."

I think that should be 'move to 64-bit' or 'move from 32-bit'.

Another one

I also have to use a Windows machine at work, and access the sites as soon as the RSS gets me the news. At home I have this Kubuntu 9.10.

Well, I'm being counted a few different times ;-)

I use Windows XP, Windows Vista and now Windows 7 to access TuxRadar regularly. I also use Ubuntu 9.04 and now Ubuntu 9.10 and Mac OS X 10.4 (Tiger) and soon Mac OS X 10.6 Snow Leopard. That's because I use all of these desktops in my work environment.

That should cause enough confusion in the stats LOL

Regards from

Another Work Reader

I also read from work on the barely functional and expensive malware known as Microsoft Windows *yuck*

I am however, reading from home atm on my new shiny Mandriva 2010 distro :)

More linux! More!

I think someone needs to DDOS (not to the point of denial though) tuxradar with MikeOS ARM lynx packets...


Oh wait a sec, MikeOs has no networking support!

That's IT!

I'm off to write that TCP/IP stack for MikeOS tonight.
or maybe not.. but it was a good idea though ;)

64 bit

Well, of course most of Linux users still use 32 bit version, because you don't need the latest hardware to comfortably run Linux, I use a 5 years old celeron, the cheapest I could buy at a time, and it still runs great. I never did a fresh reinstall either, just updates and upgrades on my Debian testing distro.

64 bit vs 32bit computers

I agree with Eric that it would be even more interesting to see the percentages of the 32bit linux users who run their distro on a 64 bit machine. Certainly people who does not have a 64bit processor are not able to run the 64bit distro.

And BTW does not all the time linux advocates emphasis that how superior is linux (vs. Windows) because it is capable to run with a decent speed on older machines??

At home:
-The only 64 bit machine run Ubuntu 8.04 32bit.
-My 32bit laptop run Mint Linux 7 32bit.
At work:
-Computer1: Ubuntu 9.04 32bit run on a 64bit computer (company support reasons)
-Computer2: Linux Mint 7 64 bit

False Colours?

I noted that bit about 'not all distributions identifying themselves'; does that mean they don't display anything or just show a distribution they derive from? So, could my 64 bit Mint 7 be pretending to be Ubuntu instead?

@tux radar

I have to agree, this release is outstanding. Are you guys going to talk about it on next week's podcast?

Not all Linux machines appear as Linux

If you OS fingerprint the IP address this posting appears to come from, you'd think I was on Windows 2003. However, that is a firewall at work (no I don't know why we have a Windows firewall - to me that's like using ice cream as a heat shield but I digress). This posting is coming from a Fedora 11 laptop.

@Ray Woods

Your 64 bit Mint 7 does show up as Ubuntu 9.04.

Linux is in many cases the

Linux is in many cases the only current OS that will run on older machines. Therefore, the higher percentage of x86 versions is not surprising at all.

The 64-Bit Issue

The only reason I'm running a 64-bit version of Ubuntu is that my new desktop machine has more then 4 GB of memory, and I wanted to make use of the extra (I often run multiple virtual machines at once). While Ubuntu-64 is, generally speaking, able to run just about everything I need, there are still one or two trouble spots. Until those areas are taken care of and everything works out of the box, I'll be tempted by the prospect of going back to a 32-bit distro. I can just imagine most people sticking with 32-bit just to avoid the hassle.

You say that the number of

You say that the number of 64bit Linux users is moving at a glacial pace. I think you're kind of missing part of the puzzle here; Linux systems actually have the potential to run on much older hardware; much older hardware is not 64bit, so perhaps what you are seeing are a large % of 64bit users which are greatly outshadowed by the number of users with old hardware.

Put a poll on your site

You should put a small survey on your site asking Windows users where they are browsing from (work, home, etc..) And also ask if at Work they have the option of using Linux. Most will probably say No.

Then you can apply those numbers to the ones in this article to discount the many Linux users who are "forced" to use Windows.

About 32-bit and 64-bit

"We're really hoping that Ubuntu 10.04 will make 64-bit the standard choice when downloading"

I'm hoping they provide the right page to the user if it can detect whether the system is 32-bit or 64-bit from the browser data. That would be the most convenient to most users.

Why I use 32-bit

I bought a Belkin wireless N card (having had good experience with Belkin and drivers in the past) and it comes with some really obscure chipset that has no driver (and doesn't really have a hope in hell of ever having one). Therefore I had to use ndiswrapper. Now, ndiswrapper is a little picky about 64-bit drivers; it will only accept ones for 64-bit Windows XP. And (surprise) there are no drivers for 64-bit Windows XP. So I had to switch to 32-bit.

If any budding wireless driver writers happen to be on tuxradar, please get hold of and write a driver for the chipset:

Ethernet controller [0200]: Marvell Technology Group Ltd. 88W8361 [TopDog] 802.11n Wireless [11ab:2a02] (rev 03)

(OK, I don't really see this happening, but it's worth a try, no? :p)

The twenty percent would have to come from someone.

<blockquote>As with last time, even if our Linux users increased by 50%, they still wouldn't match the Windows number</blockquote>

Well no. The twenty percent would have to come from someone.

Why use 64 bit.

So really what advantages are there to using 64 bit linux?

Lot's of that proprietary stuff that is nice to have ( flash, codecs... ) cannot run in it.


I use 32-bit because I'm lazy. I could swap to 64-bit, but I don't.

Using a windows machine because...

I'm at work. That's why here too.

@BubbaT: Yes it can, and the

@BubbaT: Yes it can, and the 'Buntus do a very good job at making it completely transparent; any 32-bit app will run. And I haven't noticed any codec problems. I was using 64-bit Kubuntu 8.10 and later 9.04 for more than half a year and had no problems other than the ndiswrapper thing.


And as for the advantages, you might notice a slight speed increase in all areas, but the most dramatically increased area is by far media-related conversion and playback - this is not in relation to 64-bit itself, but rather the media extensions - most 32-bit processors towards the end of their lives were coming with loads of media extensions but most apps weren't compatible with them as they had to be compatible with processors without these. Every 64-bit processor is guaranteed to have many of these extensions, so the apps are compiled with this in mind and you get a much-wanted speed increase.

This main advantage does not apply to Windows as people generally compile all their apps for 32-bit (still).

Visiting with Linux

I am visiting today using Ubuntu 9.10 on my flash drive at work. Ha ha, I RDP'd to my remote Windows desktop so I can keep tabs on the queue and my MS Outlook. Yes I could also hit it via the web but we do have one proprietary app that will not run under Linux. BTW using the 8.04 version I was unable to connect this way. I tried but no success. Linux rules.

Good stuff!

Thanks for the numbers.

Very Good Post thanks for the info - 64bit is self solving.

Linux x86_64; Ubuntu/9.10 (karmic)
No problems with 64bit.

All new PC's are 64bit so we just need to wait a while.
Also as ppl buy more ram they will move of their own accord.

When I moved to Ubuntu I had 4GB of RAM and decided to go 64Bit, never regretted it.

I now have ungraded to 8GB and as more do the same the will move over naturally.

re the usage figures very surprised by fedora only getting 10%

Hmph, I dunno. I use Windows

Hmph, I dunno. I use Windows on my desktop being a cheaper alternative to a Mac desktop publishing lab.

Although my two servers running Slackware and FreeBSD are nearly error and problem free for over 4 years running, I don't have them reporting to this site. ;P

My daughter on the other hand, wow. She's all about Debian, and she's only 9!

Sorry, I can't help it.

I'm one of those tirds looking at the site from Windows. Alas, i'm all Linux at home, but at work the only option in XP. Since i have more free time at work than at home, i rarely come here at home. I wonder how many others are in that boat? Why else would you come here?

who uses what when where ?

I work for a major public service in the Netherlands, where they so-called advocate ODF but let themselves be bribed by M$, for hundreds of millions ! So everything that was, is and shall be is M$. And yes, I do surf around on the job, between two cases, it gives me a mental break ;-)
At home, I have to maintain WinXP for other than personal reasons. I especially have to find out how to protect my wife and especially my daughter from all the evil out there :-(
And when that is done, I switch to my other pc, Mint7 dual wide screen, runs perfect, my work horse. Mint7 will identify itself as Ubuntu, and I think that might distort your statistics a bit.

Lies, damn lies and statistics...

"The fact that Windows 7 appears in three of the top four articles on our site should come as no surprise ... Of course, it's also clear that Ubuntu also appears in four of the top six"

Isn't that more of a insight into what you write articles about rather than which articles your readers find interesting?

It's very unlikely that Slackware, or Gentoo, or Arch could ever appear in four of the top six considering the small number of articles you've written about them.

Don't get me wrong, not complaining here, just pointing out that your results are skewed by your own content choices which are for the most part heavily Ubuntu focused.

For the record: OpenSUSE 11.2/KDE

@TuxRadar: Folks explain why they (have to) use Windows.


> See the comments on the previous story - a lot of folks took the time to explain why they (have to) use Windows.

Could you please point me to that story? Is it one of the six the article mentions? I did search some article comments, but did not find what I was looking for..

Thanks in advance, markus

PCLinuxOS is a rolling-update distro

'PCLinuxOS hasn't even had an official 2010 release, and yet its users have already installed the latest packages to effectively bring themselves to where the 2010 release would be - now that shows where the really eager Linux users are.'

I think that this says more about the nature of PCLinuxOS itself than the preferences or behaviour of its users.

The nature of its rolling-update system means that partial updates of your system are not practical. If you have to update one package you are very strongly advised to update your system 'across the board', and failure to do this can lead to problems. This means that all users of this distro who are updating their boxes at all will be getting the latest and greatest versions of everything that they have installed.

The irony of this is that many users of this distro (including me!) are actually most strongly committed to it on account of its quality control and its cautious release schedule. We do not all choose it because of a hunger for bleeding edge software and features.

comments on comments

Lots of good comments here.

@bout content, it seems true, this Tuxradar seems a bit biased toward Ubuntu, but why shouldn't it be? At some point people have to cater to the majority (of the minority), particularly if we're going to sway the majority (as in the majority) to switch away from proprietary software. The point is, I as a long time linux user find most people have a hard time switching because quite honestly, they struggle to really even know what 'linux' is. Mark Shuttleworth and Canonical have done wonders toward 'mainstreaming' linux, despite flaws. I think the momentum for Ubuntu is there, and speaking for myself, I'm happy to roll with it.

@Davince and others concerned about the high/growing percentage of Windows users. I think it is a good sign there is such interest. Ok, yes the "from work" issue biases things, however, I think it also shows (as with Leopaard increasing) that people are wanting to know more about linux (maybe getting tired of their OS finally?? Realizing there is an alternative??)

I use it 100%, aside from a really stupid Cisco VPN dependency on Firefox 32bit libraries for certificate management, I can say it seems excellent. No problems with flash or adobe in general. And a terrific benefit when we talking about RAM usage and efficiency.

Thanks Tuxradar for the figures. Interesting bits for a quick analysis. By the way, to look at everything together, have you tried a log scale?

Not always to newest distro

I used Hardy Heron from june 2008 to early Nov 2009. Didn't upgrade it to Ibex or Jaunty. I moved to Karmic coz Linux Mint Gloria was my default distro since june 2009. Mint is still the default OS to my children and my wife but i'm mostly using Karmic or more likely Mandriva 2010 One KDE.

I gotta feeling that average Linux-user has at least 2 distros and perhaps another partition for Windows OS on their computer. It would surprise me if there is even 10 million computers with Fedora on it. But there must be more than 15 million with Ubuntu. Ubuntu and Mint are propably those for newbies and i don't wanna any of newbies to jump very early to Mandriva, Fedora or the others.

So try Ubuntu/Mint first and then later try Fedora/Mandriva or some others...

You can run 32bit Firefox in 64bit Linux

At home and work I run 32bit Firefox (official from Mozilla) and 32bit mplayer on 64bit Linux. You can uses all codec + plugins but have to stay with 32bit browsers and muedia players.

Using crtappy XP with IE6 now while o0n holiday. The CDROM drive broke so I cnnot boot my puppy linux.

Using Linux Mint 8 (Helena)

What does Linux Mint show up as in your stats...Ubuntu or Mint??
Great site BTW...lots & lots of AMAZING articles and tips. I rekon 2010 is going to be a huge year for Linux with all new updates and better desktops maturing. I've already starting pushing Linux Mint down here in East Africa (Kenya) big time via my LiveUSB demo. People here LOVE it!!
We just need to find a way to run Quick Books so these addicts can happily move away from Virus-prone Windoze.

Go Linux (Mint) go! :-)

Max (aka Max "The IT Pro")

PS---> Am writing this in Safari on Mac OS X - a friend's desktop

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Username:   Password: