Most servers around the world run Linux. The same goes for almost all supercomputers. That’s astonishing in a capitalist world where absolutely everything is commodified. Why can’t these big tech companies manage to sell their own software to server operators or supercomputers? Why is an open, free project that is free for users so superior here?
True.
Lots of companies large and small are running commercial distributions of Linux including with paid licensing for products Like RHEL (Red Hat Enterprise Linux) or SUSE. Millions of other servers are on co-branded versions of Linux that are provided for free to the customer as long as the customer continues to use the company’s service. Examples here are Amazon Linux or Oracle Linux (both of which you’re only allowed to run if they’re operating on Amazon or Oracle Cloud servers. Now, these same companies likely also use unlicensed free Linux in places, either disposable clusters or labs, but if an application has commercial uptime requirements (meaning downtime costs money), few companies run free Linux in those specific applications.
This is a frighteningly small install base to try to sell a commercial operating system on. How many supercomputers are there in the world? Perhaps 1000? Moreover, these are such specialized set ups that trying to make a one-size-fits-all OS is likely impossible.
Just because there is free Linux does not mean that all Linux servers in the world are free.
I’m pretty sure you can run Oracle Linux on bare metal? But it only makes sense if you plan to run Oracle software on it (they only support enterprise distros like Oracle Linux, RHEL, or SLES) or want to use Ksplice to patch the kernel without rebooting.
Yep, you’re exactly right. I was incorrect on that. I’d never looked at trying to license it on bare metal (or for use in another Cloud provider), but I looked it up and found you’re correct! I had assumed it was following the same licensing model as AWS does with Amazon Linux.
Nobody runs AWS in their data center, but lots of people have a humongous and ancient oracle database or two running. Oracle Linux was forked from RHEL in the mid 2000s for this use-case.
I never had any interest in it because it didn’t make sense to run Oracle Linux for the DB and some other distro on everything else, so we went with a more mainstream enterprise distro we could use for everything.
After they acquired Ksplice and ruined it for everyone else they have a better value proposition for it, since now they’re the only ones who can patch kernel vulnerabilities without rebooting.
Well, that’s exactly what the AWS Outpost product does. Oracle has the same type of product for OCI called Cloud@Customer. This is on-prem equipment that runs the cloud vendor’s hypervisor along with integration into CSP.
While you’re right about its origins, and options for bare metal use, in addition to that Oracle has evolved it to be their “free but supported” enterprise grade Linux for VMs running in OCI, just as AWS does with Amazon Linux.
I completely agree with your approach for on-prem deployments. However, for OCI VMs its a compelling case to use Oracle Linux when there’s no licensing costs compared to RHEL or SLES while still being an Enterprise supported OS.