CentOS announces reduced lifecycle on CentOS 8 and Stream Focus

Submitted by Scott Dowdle on Tue, 12/08/2020 - 13:49

Regarding CentOS Project shifts focus to CentOS Stream:

My following statements will pigeon-hole both Fedora and CentOS as being a-certain-thing when they are really nuanced and multi-faceted. Fedora is way ahead of RHEL... and RHEL was usually freezing on a version of Fedora and then building on it for a year to a year and a half before it became RHEL and by that time, Fedora had kept on moving with 3 more releases. So while Fedora is (again in a single aspect) the proving ground for new technologies... it led by alot.

CentOS has always been a lagging follower of RHEL... as it rebuilt from RHEL sources and it usually took them time to figure out the build requirements to get as close as possible... as again... RHEL building a distro over 12+ months... different packages had different build requirements because they were made at different periods of time... and figuring those out, in a few cases, can be difficult.

So Red Hat has had a dilemma (from my perspective, I'm not sure if they acknowledge this as being an issue or something by-design) over the last couple of releases.... they have taken longer and longer to have a new major release. I think part of that is because of the large number of changes that happen in Fedora so quickly.

By creating CentOS Stream, RHEL wants to shift from being based on Fedora to being based on CentOS which puts CentOS in-front of RHEL rather than behind... but not very far in front. Fedora still plays a role in being where all of the new stuff gets developed and battle-tested before being integrated into CentOS Stream and then being integrated into RHEL.

My guess is that CentOS Stream 8 won't be that radical and that it will serve the needs of most folks who would like to run RHEL but don't want to pay for it... so realistically, I don't think much is going to change from a user's perspective.

There has ALWAYS been the feeling that RHEL and clones were just too old to run once it was several years into its lifecycle. I mean, out of the gate, it is a 1.5+ year old Fedora release (loosely).  There has always been a significant number of people who wanted fresher stuff in EL and with CentOS Stream 8  they should (maybe) actually be happier... although RH has addressed quite a bit of that issue with AppStreams (aka "modularity").

I also got the perception, although I can't really point to any specific data to support it, that CentOS was overtaxed trying to support BOTH CentOS 8 and CentOS 8 Stream.  Another perception was that EL8 just hasn't had the uptake of previous new releases... and more folks are sticking with EL7 for longer, again without any real data to prove that... but I think it is a fairly common perception on the #centos channel on their Freenode IRC channel.

SL (Scientific Linux) decided NOT to do EL8 because they said CentOS was doing such a good job that they didn't need to duplicate the effort. This announcement changes that, so maybe SL will change its mind. If they don't there is also Oracle EL, which they give away for free, and you only have to pay for it if you want support... although I haven't looked at their terms. If no existing related EL clone picks up the slack, surely one or more communities will spring up to do so. It will be routed around if needed.

Red Hat has also discovered it quite hard to explain the relationship between Fedora, CentOS and RHEL (a Penrose Triangle?)... and I think this change actually alters the relationships in such a way that makes it much more understandable.

Of course there are those who are crying foul... often folks who have called many fouls in the past, many of which have turned out to be inconsequential.  We'll have to see how this pans out, but I'm actually optimistic and think the change can potentially lead to a few unanticipated benefits.  Will I have to eat my words?

For now though, a big THANK YOU to Red Hat and the members of the CentOS community for all of the hard work they have put in in the past and what they'll be doing in the future.  I often tell people I run Red Hat-based distros and some tell me they do not... and I'll take the opportunity to comment on how Red Hat contributes so much to the mainline Linux kernel every release cycle (~9-12% or so) as well as sponsors dozens and dozens of core FOSS projects that are used by everyone... that (almost) everyone who runs Linux, is using Red Hat... just perhaps indirectly.