Regardless of the name of this blog and my current employer, I’m still a
UNIX guy first and foremost, and as such I can’t help but ponder the implications of this week’s little IBM/RedHat drama. And rather than just link to it with a short quip, I think a little more is in order.
I used to run Red Hat Linux way back (I shunned the chaos of Slackware, and still have a
4.3 CD someplace as a memento of my switch over), and many of my telco and financial sector customers use it (largely due to the enterprise support, ecosystem, and now, OpenShift). And, of course, Oracle has its own spin on it (now “unbreakably” obfuscated under their product line), so it’s effectively inescapable wherever you go.
Also, a few of my friends have (maybe a little stubbornly) stuck to CentOS for decades because it was stable and predictable. And, for a long time, I agreed with them. I spent way too long dealing with RHEL variants (like the Cobalt boxes, for which I backported untold amounts of RPMs) and knew most packaging intricacies (and kernel limitations) inside and out.
That lasted until I found myself in a Debian-centric org and realized there ought to be something a little better than either, at least where running software that had actually shipped in the same year I got my machines was concerned (both Debian and CentOS were always umpteen kernel versions behind barely working on any of the hardware I had, and using LXC and Docker remained a major pain for several releases).
Which is why I’ve been using Ubuntu for around ten years now – great hardware support, modern packages, and you can almost forgive them for their default desktop because there were plenty of variants – and were it not for hardware rotation, I’d probably still have a machine that had gone through four or five direct, seamless LTS upgrade cycles (right now pretty much all my machines (Intel, ARM, virtual or not), are at
20.04, except for the ones running Elementary, which are just fine with
18.04, and a few Pis). My experience has been that as long as you never use
.10 releases on servers, you’re golden.
Having Ubuntu as the default on WSL was serendipitous, and means I have a little white-on-black island of sanity tucked away inside any of my Windows 10 machines
So the fallout from CentOS Stream becoming what is essentially RHEL “unstable” and the knee-jerk emergence of Rocky Linux as the new downstream de-branding (de-fanging?) effort are something I look bemusedly upon, even as I cheer on Ubuntu (which has recently gone “pro” and is taking on a more enterprise stance).
And yet, maybe we should count ourselves lucky, in some regards. After all, it’s been two years since IBM bought Red Hat and things were rather quiet. I always wondered if IBM would just sit there and continue to allow people to effectively run RHEL (under the guise of CentOS) without licensing fees, and now we know the answer.