I can still remember my first experience with Linux. I was an intern at IBM in 2001 and downloaded the IBM Linux Client for e-Business onto a 3.5″ floppy and set about installing it to a test machine in my cubicle. It was based on Red Hat 6.1. I had lots of fun recompiling kernels, testing broken applications (thanks Lotus Notes), and trying to get basic hardware working (thanks deCSS). I couldn’t help but think at the time that there was great potential in the software.
I’ve played with Linux on and off for the last twelve years. SuSE, Novell, Ubuntu, Gentoo, Slackware, and countless other distros too obscure to rank on Google. Each of them met needs the others didn’t. Each tried to unseat Microsoft Windows as the predominant desktop OS. Despite a range of options and configurability, they never quite hit the mark. I think every year since 2005 has been the “Year of Desktop Linux”. Yet year after year I see more Windows laptops out there and very few being offered with Linux installed from the factory. It seems as though Linux might not ever reach the point of taking over the desktop. Then I saw a chart that forced me to look at the battle in a new perspective:
Consider that Android is based on kernel version 3.4 with some Google modifications. That means it runs Linux under the hood, even if the interface doesn’t look anything like KDE or GNOME. And it’s running on millions of devices out there. Phones and tablets in the hands of consumers world wide. Linux doesn’t need to win the desktop battle any more. It’s already ahead in the war for computing dominance.
It happened not because Linux was a clearly superior alternative to Windows-based computing. It didn’t happen because users finally got fed up with horrible “every other version” nonsense from Redmond. It happened because Linux offered something Windows has never been able to give developers – flexibility.
I’ve said more than once that the inherent flexibility of Linux could be considered a detriment to desktop dominance. If you don’t like your window manager you can trade it out. Swap GNOME for xfce or KDE if you prefer something different. You can trade filesystems if you want. You can pull out pieces of just about everything whenever you desire, even the kernel. Without the mantra of forcing the user to accept what’s offered, people not only swap around at the drop of a hat but are also free to spin their own distro whenever they want. As of this writing, Ubuntu has 72 distinct projects based on the core distro. Is it a wonder why people have a hard time figuring out what to install?
Android, on the other hand, has minimal flexibility when it comes to the OS. Google lets the carriers put their own UI customizations in place, and the hacker community has spun some very interesting builds of their own. But the rank and file mobile device user isn’t going to go out and hack their way to OS nirvana. They take what’s offered and use it in their daily computing lives. Android’s development flexibility means it can be installed on a variety of hardware, from low end mobile phones to high end tablets. Microsoft has much more stringent rules for hardware running their mobile OS. Android’s licensing model is also a bit more friendly (it’s hard to beat free).
If the market is really driving toward a model of mobile devices replacing larger desktop computing, then Android may have given Linux the lead that it needs in the war for computing dominance. Linux is already the choice for appliance computing. Virtualization hypervisors other than Hyper-V are either Linux under the hood or owe much of their success to Linux. Mobile devices are dominated by Linux. Analysts were so focused on how Linux was a subpar performer when it came to workstation mindshare that they forgot to see that the other fronts in the battle were being quietly lost by Microsoft.
I’m not going to jump right out there and say that Linux is going to take over the desktop any time soon. It doesn’t have to. With the backing of Google and Android, it can quietly keep right on replacing desktop machines as they die off and mobile devices start replicating that functionality. While I spend time on my old desktop PC now, it’s mostly for game playing. The other functions that I use computers for, like email and web surfing, are slowly being replaced by mobile devices. Whether or not you realize it, Linux and *BSD make up a large majority of the devices that people use in every day computing. The hears and minds of the people were won by Linux without unseating the king of the desktop. All that remains is to see how Microsoft chooses to act. With a lead like the one Android has already in the mobile market, the war might be over before we know it.
Remember Tom, Ubuntu is based upon debian. Debian, according to distrowatch, has 140 distributions that are based upon it. This doesn’t add much to your article, but I had to post it as I am a debian zealot. 🙂
Linux “advocates” keep shifting the goal posts on what constitutes Linux. To me Linux is a full Linux distro not something that runs just the kernel or part of the GNU programs. In that case, it could be said that Linux as a kernel came out very successful (hell, even FreeBSD is very successful given all those Juniper firewalls) but Linux as an OS is a failure (from a marketshare perspective).
On the desktop it has failed.
On the mobile it has failed. See Meego/Tizen/Maemo/Ubuntuphone
On the server you have a partial victory. Linux is mostly deployed in public facing servers, not “office”-type environments where Windows and AD dominate.
I will agree with you that no single distro has succeeded in toppling the Windows monopoly. In part because it’s just too easy to create a new distro from thin air if you don’t like a package choice. Imagine if Windows had that kind of flexibility? Windows 8 would have been more successful because people couple swap out the window manager for the one in Windows 7 and get new HW support with the same old interface.
I don’t think Linux distros will ever succeed so long as there are a multitude of options out there to satisfy a given user’s whims. That’s why I chose to look at the “essence” of Linux, which was the kernel itself. To me, that *is* Linux.
Tom, buddy, you contradict yourself in a single post. The strength of Windows is in its uniformity. If you start fiddling with serious components like the WM you open the can of worms as the changes cascade across the entire OS. Pretty soon you have a mess of incompatible OSes. Testing and deployment become a nightmare. The same Shotwell has different bugs in Suse vs. Debian vs. Gentoo. Even though the upstream components are the same, you have yourself a software bedlam. One has to bend over backwards in developing for Linux because testing becomes impossible when everyone is using patched forks of upstream software. Debian is probably the worst offender but everyone does it to a degree.
In servers its the same thing. You always run stock RHEL. This is why things like OpenVZ and XEN were hated by sysadmins. You had to patch the kernel and this rendered the OS not binary compatible anymore. You were essentially f***ed when its came to support contracts once you started to fiddle with the components.
This reminds a bit of FC SANs. You never want to mix vendors because everyone diverges from the standards in the own ways and its impossible to test all the permutations.
The biggest problem for Linux in my understanding is that it doesn’t have ‘production’ repositories – in particular Debian. There wouldn’t be more than 3 packages in a ‘production’ repository I reckon. By a ‘production’ repository I mean something that is completely polished in the same way as products are for the Windows world. It is still a geeky environment where virtually everything has to be massaged to get it running. I can’t even get my screen res to stay without it downsizing when I switch the KVM !! There is a long, long way to go. I’ve been using Debian and Ubuntu for 14 years. I am sick of massaging I just want plug-and-play.
The war was lost to NIX long ago. NIX servers outnumber Windows servers and have for a long time. Total NIX devices far exceed Windows. The fact people have a hard time using NIX for desktop has nothing to do with NIX’s dominance elsewhere. What do you mean there are no “production” repositories? There are many and it seems by polished you mean Windows-like which is not the goal of NIX. Windows has many issues especially with the new model they are trying to move to but it will keep its place as a desktop OS and dwindle in all other areas.
Pingback: Rome Wasn’t Software Defined In A Day | The Networking Nerd
Pingback: Cumulus Networks Could Be The New Microsoft | The Networking Nerd
Pingback: Linux and the Quest for Underlays | The Networking Nerd