Goals

Support Kube extenders to develop Operators for delivering their services to consumers through kcp.

Why is this important? 

kcp is a Kubernetes-compatible control plane that is run as a managed service by Red Hat. kcp provides three unique capabilities to our customers and internal teams:

1. Transparent multicluster - run unmodified applications across groups of clusters, embracing a fleet mentality
2. Fleet-wide Kube API extensions - Help Kube extenders to provide their Operators as services to consumers, rather than as software added to a single cluster.
3. Seamless integration with managed cloud services
   • Customers can use #2 to pull our cloud services into their environments in a Kube-native way
   • Internal teams can use #1 to run new managed services in a self-service way while working within our SRE/ROMS and productization frameworks
   • Internal teams can use #2 to export their APIs across the customer base. Eg. Everyone gets to use Kafka APIs provided by the managed service. This gets them instant reach when they GA.

What the Operator Framework team can support on the kcp front is to help drive the “Kube API extensions” in:

• the client-side tooling/libraries for building controllers that interact with kcp APIs
• the kcp-side where client calls need to consider scoping contexts
• the technologies in kcp to easily expose customized Kube APIs to tenants'/customers' workspaces
• the technologies/SDK to enable controllers to support cross-shard communication in kcp sharding for multi-regions

Deliverables (epics)

• client-side tooling/libraries - building controllers that interact with kcp APIs
• scoping client calls in kcp - the kcp-side where client calls need to consider scoping contexts
• API catalog - the technologies in kcp to easily expose customized Kube APIs to tenants'/customers' workspaces
• multi-workspace/multi-sharding SDKs - the technologies/SDK to enable controllers to support cross-shard communication in kcp sharding for multi-regions