As we now all know, Oracle is planning their own Linux. This is causing general hand wringing and dire predictions of doom. For example, according to
(link dead, sorry)
Oracle's Unbreakable Linux 2.0 likely won't fragment the Linux distribution business out of the gate, but the long-term impact on Red Hat, Novell and the operating system market remains unclear, industry observers say.
Market fragmentation? The Linux market is already fragmented and has been for some time. Sure, the big boys have the lion's share, but even if you ignore all the little distros the fragmentation is no worse than that which supposedly plagued the kindred Unix market. If fragmentation was dangerous for Unix, it's just as much so for Linux and Oracle's move doesn't change anything significantly.
What I think is being missed here is the importance of
virtualisation. The iTnews.com article does mention that:
"ISVs need to control their application destiny as virtualisation accelerates. They have to control what OS components are installed with their application in the virtual container because of efficiency and supportability," said one high-tech venture capitalist, who declined to be named. "ISVs need to control their own stack, and it becomes increasingly important as virtualisation comes to fruition."
However, I think they (and a lot of other people) aren't seeing the long term picture. The forces pushing toward VM's are from all sides: the application vendors want it to control their environment, and the users want it to control security, for hardware deployment flexibility and to cut hardware costs. There's no stopping it: the VM world is coming.
Now right now, that world is VMWare or Xen on Linux or Windows, or Microsoft's Virtual Server. The virtualisation is separate from the OS. Oh, tightly coupled in the kernel, sure, but the point is that the OS flavor of the VM is still important: you are running the other VM's under Windows or Linux. That is going to change: the real OS (the controlling OS) will be a hypervisor like VMWare's ESX Server: a smallish, very tight, special purpose OS whose only role is to provide the framework to virtualise other OSes that users interact with.
That's why Linux and Windows and MacOSX are going to become unimportant. Not irrelevant, but more and more focused on running specific software inside VM containers: in other words, appliances. From an evolutionary perspective, consider this as a major environment change. The flora and fauna will adapt, and will lose features they no longer need.
Who has the most to gain here? Why Linux, of course. Not Oracle Linux, not RedHat Linux, not Suse, not any specific vendor, but generic Linux, modifiable Linux, malleable Linux. If anything, Linux becomes more fragmented in this environment: a Linux distro for every app isn't hard to imagine. But they won't be distros in the traditional sense. Sure the "vendors" will release their mods back into the mainstream, but a lot of those will be so specialized to their needs that no one will care: they'll never be part of the mainstream kernel. In fact, that mainstream kernel becomes almost totally unimportant. It remains the starting point from which apps build their container appliance, but that's all: by itself it will be unimportant.
There will be holdouts, of course. People so tied to Microsoft habits that they can't move. But they'll virtualise too; they'll have to. The only difference is that they'll be paying a Microsoft tax, so will have incentive to lose that dependence. Eventually they'll move to generic Linux also.
I hear you: "But that makes Linux even MORE important!", you protest. OK, yes, you can look at it that way. But remember: it's not RedHat Linux, not Suse, and not anyone else making a general purpose Linux. That environment is going to go away. Linux as a general purpose OS won't matter. Linux as part of a container application will matter.
In a larger sense, operating systems never have been important. It's apps that matter, it's apps that we use. The OS is just the scaffolding that lets our apps run. The VM world just reinforces that reality and makes it more obvious to us.
So iTnews is wrong, I think: the long term effect on the operating system market isn't unclear. The OS market is going to be destroyed outright. The operating systems themselves will die if they cannot adapt to the new environment, but Linux obviously has the most adaptibility here. Remember, that still means the Linux OS market dies - or more accurately is subsumed by the application markets. Linux won't be important to computer users, only to the application vendors who use it to power their wares.
Of course, I could be completely wrong.. if you think so, I'd love to hear why.
Got something to add? Send me email.
More Articles by Anthony Lawrence © 2012-07-02 Anthony Lawrence
Unlike info, pinfo does not display anything if it has nothing. I've been forever irritated by info coming up with its default page when it has nothing to tell me. (Tony Lawrence)