Can someone explain to me the hype around KRO? Asked on linkedin too but I’m still confused.

This feels like a baseline framework that the big cloud providers will build the next offering around. While you can design your own abstractions with it, it gets tedious fast if you want to spread them consistently to a large fleet of clusters. Who controls which of those abstractions can be accessed by which developer, aka RBAC? How often, aka quotas. Ton more of concerns that you might want to supply as a cloud console based administration tooling and distribution service, aka “platform”.

Its in my mind a poor mans crossplane in many senses. It is easier but WAY more limited and constrained. I believe it only exists because crossplane MRs are cluster scoped for now (changing in 2.0). It is very underwhelming in my mind but in the end enforces the idea that easy building of cluster side compositions of resources is a real thing, and its just another tool trying to solve this area

@Scott Rosenberg interesting re. the cluster scope - had not been aware of that

Yep. Crossplane initially went with namespaced claims and everything else cluster scoped similar to PVC and PV and with the benefit of ensuring unique names when dealing with cloud resources, and also with the notion of users not needing to deal with the underlying abstracted resources. In theory i can get the idea, but especially in multi trnant environments it becomes very challenging. Luckily this is changing soon

What Scott said, it feels like a poor man's crossplane, almost like instead of contributing upstream they decided to do it themselves, which seems to happen a lot with hyperscallers for example Microsoft has https://learn.microsoft.com/en-us/azure/azure-resource-manager/bicep/overview which is a bit like their version of terraform