Red Hat, IBM, and the Community: the M&A Bait and Switch That Doomed CentOS

5 min read

“Nothing will change.”

This statement is part of the merger and acquisition playbook for nearly every company; in reality this is impossible to be true. Any M&A activity by its very nature is a change, but this is business theater and nobody wants to make more waves then they have to, especially looking back at IBM’s multi-multi-billion dollar acquisition of the Linux King, Red Hat.

When this was announced, I actually drew up a dead pool to bet myself what IBM would do to nuke RH culture and product lines, and I have to admit while pulling support for open source products was one of the items on the list, I did not expect it to be with CentOS.

I have had kind of a love-hate relationship with CentOS for a long time, but I was more forgiving of it before it became a wholly owned Red Hat project; CentOS 8 took more than 4 months to release after the Red Hat Enterprise Linux (RHEL) 8 initial release and instead of speeding up the rebuild process, has continued to putter along, 1-3 months behind.

Of course, now we do not have to worry about that at all. IBM/Red Hat has sentenced CentOS and its community to digital death on December 31st, 2021. As Nasrudin’s fable goes, and I am paraphrasing, we must convince the King that a horse can sing to keep CentOS alive.

At first I was pissed. I have been an advocate for startups and technically savvy organizations using CentOS as a drop-in replacement for RHEL since the mid-late 2000s and by the same token, advocated to move to RHEL at a point where enterprise stability and support was a necessity for those organizations and this felt like a personal betrayal to me where I have funneled hundreds of thousands of dollars to Red Hat over the years through that exact pattern.

Some of the community has encouraged considering the justification of “Why support a project that competes with their money-making product?” And also, “It is just a change to upstream, and the rolling release pattern.”

As to the first: some people will never pay for RHEL subscriptions. That is just the way it is. Now, however, the organizations that needs stability and not rolling releases will look towards somewhere they can get a full, stable enterprise linux distribution with an LTS release - probably Debian or Ubuntu, not because they can not switch to another community rebuild but because many people, like myself, feel betrayed by Red Hat’s (now IBM’s) governance of CentOS and a bait-and-switch.

For the second item: many modern organizations are fine with a rolling release; they have no vested interest in the maintenance of the OS so long as it works and security fixes are delivered. The change from downstream to upstream is a bigger issue. That means that it becomes the place that the bugs hit first and has no real guarantee of the enterprise support fixes that RHEL subscriptions carry. Yes, some of those will be handled as part of the build cycle, but the likelihood is that RHEL will have to backport fixes to upstream packages because there is no incentive for them to give them away.

I could understand this starting with CentOS 9 - that would make perfect sense to announce it now, give everyone a few years to get used to the idea, and plan on either adopting a stream release or finding an appropriate replacement while knowing that Red Hat would live up to its community promise with CentOS 8. There would still be outrage, but we all know it would die out before RHEL9 was out the door. In fact, for all we know, IBM could dictate that RHEL9 is only a stream release itself, and then there would be very little lost at all.

InfoWorld stated that the discontinuation is no cause for outrage because we should not expect companies to give things away for free, but this too is a perceptually limited take on the issue and Matt Asay (who has a JD in Law & Technology with a focus on open source) should be ashamed of putting that in writing. Red Hat builds its entire product suite on FOSS products, adds security and maintainability to them, rebrands them as its own, and then sells “subscriptions” for the support and enablement within their product suite, because they can not legally sell most of the products as their own.

The spin on this is that readers should remember how open source friendly Red Hat is, and that is the justification for allowing them to pull the rug out from under the CentOS community. This move is not particularly open source friendly and for all we know this is the first of many similar moves dictated by the controlling interest of IBM, and even if it is just one of a few similar moves, it makes them significantly less friendly.

I could see this as part of a pattern to isolate and devalue other upstream products that did not originate inside of Red Hat, such as Ansible.

So yes, I am really upset at IBM and Red Hat. I am in the process of studying to re-certify for my Red Hat Certified Engineer (RHCE) certification, and it is really difficult to not let my anger distract me. I know, for a fact, that there are already good projects out there as rebuilds, even (and I share this with the bile forming up in my throat) Oracle Linux.

The sky is not falling, however. The Linux community is resilient and CloudLinux has stated that they will release a 1:1 downstream rebuild of RHEL and the project originator of CentOS, Greg Kurtzer has announced Rocky Linux as a replacement. CloudLinux has gone so far as stating that they will provide a straight transition from CentOS to CloudLinux with no major changes, and I would be surprised if Rocky did not do the same thing.

For me, and for many, this is part of the dark path of community projects controlled by a single entity and proves the point of things like the Cloud Native Computing Foundation and Apache Foundation existing - while there are corporate board members as part of those foundations and they have active say in the steering of the projects, it provides a layer of insulation from a single entity arbitrarily making a change and effectively telling their community that their promises do not matter.

In the end, it took about 18 months, but the blue of IBM and its willingness to gut success and a strong community for profit is starting to bleed into the Red Hat paradigm.

I think my final internal debate will be whether I, personally, pull the plug on supporting Red Hat at this point. Are they truly a trustworthy partner any longer? Do they really embody the idea of FOSS with the projects they control, such as OKD (the upstream for OpenShift) or Ansible? Can I truly be an advocate to the companies I have a responsibility to by saying “your usage of the open source version of this is safe until you are in a position to move to an enterprise supported version” or do I tell them that my conclusion is that Red Hat has proven itself untrustworthy and they should move to other products that I do not consider to be in harm’s way at this time.

I truly do not know the answer at this point. I will get re-certified this time around, and I will continue to contribute to the Red Hat and CentOS community as I can. We have until the last day of 2021 to find out, and we, the community, have a chance to teach the horse to sing. much can happen in that time: Red Hat may die. I may die. And perhaps the horse will sing.

Image of Stephen Sadowski

Stephen Sadowski

Leader focusing on quality, delivery, technical debt management, and leadership education about DevOps and SRE practices