CentOS is Ending

CentOS is ending being a repackaged but free as in beer RHEL clone to focus on CentOS Stream which will be positioned slightly ahead of RHEL releases: https://blog.centos.org/2020/12/future-is-centos-stream/

It seems there’s much end-user outcry on the mailing list about this (https://lists.centos.org/pipermail/centos/2020-December/thread.html#352168), but no one seems to be stepping up to fork what CentOS currently is and to continue developing it. There appears to be no good business model to sustain CentOS if the community won’t do it themselves (ala Debian).

It’s interesting to me that CentOS got as popular as it did. CentOS isn’t the same as RHEL and you can’t just call up Red Hat and ask to buy a support contract for CentOS (although I’m sure there’s a few small consulting shop who do offer such services). But Oracle and Canonical both DO offer to give you a free as in beer Linux distro with long term updates support and if you need it to easily buy a support contract for the distro you already have installed.

Why was CentOS so popular? Why hasn’t Oracle’s offering been more popular? Why don’t people appear to pay Canonical for support contracts?

Very interesting. I run Ubuntu LTS on my cloud servers (but I suspect I would be just as happy with Debian), and Arch Linux on my workstations, so not real plugged into the RedHat world. One of my customers likes CentOS a lot, but I can’t see any advantage of it over Ubuntu LTS (or Debian) for my server needs. Arch Linux has been nearly flawless for my workstation needs. For me, Linux has become a commodity – something that has just faded into the background and I don’t give much thought these days. So, I don’t really know the answers to your questions.

Perhaps the future is something different like NixOS or Oasis Linux.

centos is/was exactly RHEL but without support. It was popular because it did not have restrictions like RHEL that you cant use RHEL for more than one machine without buying a license. So if someone is setting up a new IT infra for some company or education system, RHEL was never a good option but the technology definitely is. So centOS filled that gap nicely. however it was purely a user consumption model, there was no easy way to fix something and feed it back to RHEL, with centos stream that will work nicely, When redhat got centos back in 2015, they never promised they will carry on centos line as it is. I think it was eating enough into the market that some bean counters think should be monetized now. So they will get initial conversions and perhaps some increase in revenue but its a long term void that will be filled by something else and surely it wont be RHEL technology, So some folks will go to debian and god forbit Oracle linux or ubuntu. CentOS founder has shown interest in doing something similar again so I would wait and watch.

To me it looks like RH has killed competition, they could have waited till centos 9 perhaps

I assume RHEL is focused mostly on large, enterprise customers (I’ve never heard of anyone using it, but most of my customers are small verticals). What makes RHEL good technology?

I’ve traditionally run Ubuntu TLS on servers I manage (perhaps a dozen cloud instances), as it has decently long maintenance window, but going forward, I might just go with Debian. At my scale, I’ll never be a paying RHEL or Ubuntu customer, so I may as well use products from sustainable OSS projects. Well run OSS projects can be very nimble and responsive. For instance, Chris filed a security bug he found in Arch Linux, and it was fixed in less than a day:

Redhat engineering is one of best when it comes to opensource and Linux operating systems, they employ most of maintainers and star engineers and let them work on their expert subjects/projects. They have come up with many technologies e.g. rpm, systemd to name a few core and diffucult ones, rpm in the past which is one of core technologies for binary distros. and RHEL is a very stable and robust platform which is supported for a lot longer than normal distributions were. So centos was offering a enterprise class platform but no support on top and thats good enough for lot of things, as centos use has proven over years. Other option is fedora but fedora moved to fast and does not have long life cycles.

I think for a lot of usecases, debian is good enough, and that speaks to its large user base. OS vendors is a dying usecase as business, the value chain has moved up. So there is consolidation and the recent centos developments are proving this. There is not enough of pie to grow, its pretty commoditized now than it was 20 years ago. So an opensource solution like debian is what most folks will turn to, and adapt it to their needs and feedback to debian the fixes they do, big players like google, fb, amzn are using this path. It given them most flexibility, and cloud spread means lot of local IT is moved away from premises for less tech businesses, and advent of flatpak, snaps mean that OSes wont matter in application space either. So really OS layer is becoming as boring as it could be.

Why is Oracle Linux so bad? I’ve heard this opinion many times, it seems no one likes Oracle.

I’ve never been an Oracle customer or really had any interactions with them so I’m genuinely curious.

the intention as it seem to me of oracle Linux is to create a centos based version that is basically to sell their cloud services to you. Its instrument to sell you more cloud services. They have not shown interest for it to be for the public good. Again its my personal opinion and I could be wrong.