Why do you need CobaltCore (or OpenStack) on top of IronCore?
IronCore's compute interface is designed to be ephemeral. There is, for example, no support for VM live migration or graphical console access planned. With IronCore, new workloads are encouraged to embrace cloud-native resilience characteristics. Heritage workloads often do not have the choice to re-platform. CobaltCore builds on IronCore's bare metal and network automation and adds an OpenStack-compatible API surface, bridging those obligations while preserving full backward compatibility.
What are the three components of IronCore?
IronCore is structured in three integrated components. Bare Metal Management provides a comprehensive Kubernetes-native solution for the full server lifecycle, covering automatic discovery and registration, OS provisioning via Ignition, declarative Day-2 Operations for BIOS and firmware, and Kubernetes support via Cluster API and Gardener. Network Automation provides a Kubernetes-based framework for automating the deployment, configuration, and monitoring of network devices, supporting different vendors through dedicated operators. Infrastructure as a Service exposes a unified declarative API for managing compute, storage, and networking resources using modular, pluggable providers designed for hybrid and edge deployments, with Ceph as the default storage backend.
When should I use IronCore directly, and when CobaltCore?
Use IronCore directly when your workloads are cloud-native and can be designed around ephemeral infrastructure and Kubernetes-native resilience patterns. IronCore integrates directly with Gardener for dynamic provisioning of Kubernetes clusters on bare metal. Use CobaltCore when you need to support existing workloads that depend on OpenStack-compatible APIs, VM live migration, graphical console access, or traditional IaaS interfaces. Both can run on the same IronCore bare metal and network automation foundation, so the choice is about the API surface and operational model, not necessarily the underlying hardware management.
I have a heritage workload. ApeiroRA does not seem to support my workload, why?
ApeiroRA prioritizes modern, cloud-native, immutable, anything-can-fail, up to shared-nothing principles for workloads and services. This enables a dynamic (e.g. software and data swarming) proliferation of services between providers in the cloud-edge continuum. But you can of course setup your heritage workload and connect to the continuum using OpenStack as part of ApeiroRA as well (without the benefits and characteristics of cloud-native resilience).
I need special hardware, or have a requirement which ApeiroRA does not seem to support?
Please collaborate with ApeiroRA (and the particular open-source subproject) and add your requirements (preferably with code contributions; no one will implement your use case unless you provide the energy). With ApeiroRA, we aim to distribute your requirements in the reference baseline, such that your workloads later can be deployed anywhere in the continuum.