It's perhaps human nature to cling to the familiar, it's the availability heuristic (we tend to reach for solutions which are easily recalled from memory- regardless of whether it happens to be the best fit.

Good enough is sometimes what you need. Case in point, a proxy 'envoy proxy', is slower than many traditional (Ha Proxy) yet is seeing adoption. Why? Because the way we operate systems has changed. Note I say systems not software. We are operating systems not software anymore, and perhaps always have- but the difference matters more now, there's just no getting around it, you're building a distributed system.

But eventually, things are truly better. Or rather, something which was seen as dependable, and stable is no longer stable nor dependable.

Increasingly Centos falls (has fallen) into this category. This is nothing about the quality or rigor of the project itself - no, it's the intangible ecosystems , social contracts which surround it. Yes, I'm lightly referring to the recent Centos rocky linux fiasco but it's the model that has fallen out of touch- open source ecosystems often benefit from fiasco's long term anyway. As an ecosystem so, Centos is less and less tied in to public cloud ecosystems, and RedHat generally is confusing to consume since the operating system isn't in the thought path when thinking about distributed systems, it's in the way. It needs to be there of course- it just isn't the focus when thinking in terms of a distributed system. At a minimum think in threes not single units.

Moving away from managing and thinking in terms of individual machines moves the responsibility of managing operating systems elsewhere (cattle not pets). Emphases on elsewhere, we still need and value operating systems development- of course we do, but the focus has changed. We must think in terms of distributed systems, not individual machines nor providers.