Hyperscalers Have Been Making CentOS 9 Stream More Attractive With New Features

[mether]

If they are having to backport things to it, then isn't it too outdated for their purposes?

You can't have the very latest components and preserve long term stability for 10+ years at the same time easily. So there is some tradeoffs for using layered optional repositories.

Red Hat wants to maintain a stable ABI for its enterprise customers. A number of their customers depend on third party vendor software that will break otherwise. Some of CentOS stream users want to leverage CentOS but tradeoff some ABI stability to selectively backport some leading edge components into a separate repository so they can use it within their environment.

What you prefer depends on how you are using these systems in production. If you are in heavily containerized environments or running software in highly redundant clusters, you may prefer to use the optional repository with newer versions for example. If that brings you say, better performance or some leading edge features but slightly less stability, that's ok because their systems are cattle, not pets.

[anarki2]

Yeah, no thanks. CentOS died the moment they announced the "streaming" and the C8 EOL. You either want stable or cutting-edge. The current premise of CentOS is pretty much a paradox, a mish-mash of both. If I want something rock solid I'll go for Alma or Rocky. If I want something fresh, I'll use Fedora. For the record, I switched my servers to Fedora and they've been rock solid so far. Already did 2 major upgrades (33->34->35), all smooth sailing so far. So come to think of it, I'm pretty grateful for CentOS to be a d*ck, because I'm no longer stuck with their ancient versions.

Yeah, you have those "streams", which introduce various new dependency issues, so not a big fan, even if I put the other problems aside.

I don't understand how they think anyone will willingly stick with CentOS after they signaled that "we can do anything we want". Yes, you can. And we can dump you, too.

[NateHubbard]

You can't have the very latest components and preserve long term stability for 10+ years at the same time easily. So there is some tradeoffs for using layered optional repositories.

Red Hat wants to maintain a stable ABI for its enterprise customers. A number of their customers depend on third party vendor software that will break otherwise. Some of CentOS stream users want to leverage CentOS but tradeoff some ABI stability to selectively backport some leading edge components into a separate repository so they can use it within their environment.

I guess I was mislead by the name "Stream" which makes me think it's going to be more like a rolling release, not just another frozen, soon to be outdated (but stable!) distro.

My mistake for not keeping up with whatever RedHat is doing to it.

[mangeek]

...the name "Stream" which makes me think it's going to be more like a rolling release...

The impression I got was that instead of CentOS trailing behind RHEL, CentOS Stream would lead slightly (with a tad more flexibility) before being branched off to become the RHEL point releases.

I think it makes a lot of sense, actually. You can run your desktops and DEV on CentOS Stream, and put your PROD systems on RHEL with reasonable assurance that they're close enough in lineage for things to work, and that if you have an issue in CentOS Stream, there's enough flexibility and visibility that the fix will get baked into the next RHEL.

[ehansin]

I think it makes a lot of sense, actually. You can run your desktops and DEV on CentOS Stream, and put your PROD systems on RHEL with reasonable assurance that they're close enough in lineage for things to work, and that if you have an issue in CentOS Stream, there's enough flexibility and visibility that the fix will get baked into the next RHEL.

I feel this is a good take. I figure one can be disappointed, frustrated, etc. on the change and using the CentOS name in this capacity, etc. But as a basic concept, having a distro with a more rolling/flexible nature that sits "above" the eventual stable release, doing so with attempts at ABI compatibility, has some value on its own. At least I can see that. Will add, I understand where people are at that don't like the change. Just saying that all aside, I can see value in the basic concept.