CentOS Stream 9 was released this week, marking the next chapter in what has been one of the most contentious decisions in recent open-source history. If you’ve been managing Linux servers for any length of time, you’ve likely had opinions about this since Red Hat announced the shift from traditional CentOS to CentOS Stream in December 2020.
I’ve run CentOS in production environments since version 4. The announcement that CentOS 8 would reach end-of-life on December 31, 2021 — years earlier than expected — forced a lot of us to rethink our infrastructure strategies. With Stream 9 now available, it’s worth examining where things stand.
What CentOS Stream 9 Actually Is#
For those who haven’t followed the saga, here’s the short version: traditional CentOS was a downstream rebuild of Red Hat Enterprise Linux (RHEL). RHEL would release, and CentOS would rebuild it from the source RPMs, giving you a binary-compatible, free alternative. CentOS Stream flips this relationship — it sits upstream of RHEL, serving as a rolling preview of the next RHEL minor release.
CentOS Stream 9 tracks what will become RHEL 9, which is based on Fedora 34. This means:
- Kernel 5.14 with significant improvements to cgroups v2, BPF, and io_uring
- GCC 11 as the default compiler, with C++17 fully supported
- Python 3.9 as the system Python
- OpenSSL 3.0, which is a major upgrade with implications for any TLS-dependent workload
- Podman 4.x previews, continuing Red Hat’s push away from Docker
The technical foundations are solid. The question has always been whether the Stream model is suitable for production workloads.
The Trust Problem#
The backlash against the CentOS Stream shift was never really about the technical merits. It was about trust and expectations. Thousands of organisations built their infrastructure on the understanding that CentOS was a stable, RHEL-compatible platform with a predictable lifecycle. Changing that social contract — especially accelerating CentOS 8’s EOL — felt like a betrayal.
Red Hat’s argument is that CentOS Stream is more useful, not less. By contributing to Stream, you directly influence what goes into RHEL. Bugs you find and report in Stream get fixed before they reach the enterprise release. In theory, this is a better model for the community.
In practice, many sysadmins and infrastructure teams need stability guarantees, not influence over upstream. When your job is keeping production systems running, “rolling preview of the next minor release” is not reassuring language.
The Alternatives Have Matured#
The silver lining of the CentOS saga is that it catalysed the creation of genuine alternatives:
AlmaLinux, backed by CloudLinux, released its 8.x line in March 2021 and has been delivering point releases reliably. They’ve been transparent about their build process and governance, and they’ve attracted significant community support.
Rocky Linux, founded by CentOS co-creator Gregory Kurtzer, had its first stable release in June 2021. Rocky’s pitch is explicitly “what CentOS used to be” — a downstream RHEL rebuild with long-term stability.
Both projects are now established enough that they’re viable for production use. I’ve been testing AlmaLinux 8.5 in staging environments, and the compatibility with RHEL has been flawless so far. The migration path from CentOS 8 is straightforward — both projects provide conversion scripts that handle the transition in place.
Oracle Linux also deserves mention. It’s been around for years as a RHEL rebuild, and Oracle has been using the CentOS upheaval to promote it. The “Unbreakable Enterprise Kernel” is genuinely good, though Oracle’s reputation in the open-source community gives many people pause.
What This Means for Your Infrastructure#
If you’re still running CentOS 8, the clock is ticking. December 31, 2021, is just six weeks away, and after that, you stop receiving security updates. Here’s my pragmatic advice:
For existing CentOS 8 systems: Migrate to AlmaLinux or Rocky Linux. Both offer in-place migration tools. Test thoroughly in staging first, but the process is well-documented and well-tested by now.
For new deployments: If you need RHEL compatibility, choose AlmaLinux or Rocky based on your preference. If you’re open to a different approach, consider whether you actually need an enterprise Linux distribution at all. For containerised workloads, minimalist base images (Alpine, distroless) often make more sense.
For CentOS Stream: It has a legitimate place in development and testing environments where you want early access to what’s coming in the next RHEL release. I wouldn’t run it for customer-facing production workloads yet, but for CI/CD environments and internal tooling, it’s perfectly reasonable.
My Take#
The CentOS Stream transition has been handled poorly from a communications perspective, but the resulting ecosystem might actually be healthier than what we had before. Instead of one community rebuild that everyone depended on, we now have multiple well-funded alternatives with different governance models.
Competition and choice are good for the enterprise Linux ecosystem. AlmaLinux and Rocky Linux have both demonstrated they can deliver timely, compatible rebuilds. CentOS Stream serves a different purpose — one that’s genuinely useful for developers and contributors even if it’s not what production sysadmins wanted.
What I find most encouraging is the speed at which the community responded. Within months of Red Hat’s announcement, we had multiple viable alternatives. That’s the open-source ecosystem working exactly as it should — when one path closes, the community builds new ones.
The enterprise Linux landscape is more fragmented now than it’s been in years. But fragmented doesn’t mean broken. It means we have options. And after depending on a single free RHEL rebuild for nearly two decades, having options feels like progress.