RWS Architecture article

Clusters as a Service Offering

Clusters as a Service is the first offering modeled with the DC 3.0 capability roadmap pattern. It defines the consumer-facing OpenShift cluster service separately from the platfor

  1. Typeoffering
  2. Statusdraft
  3. Domainplatform
On this page
  1. Summary
  2. Service Definition
  3. Roadmap Model
  4. Dependencies
  5. Open Questions

Summary

Clusters as a Service is the first offering modeled with the DC 3.0 capability roadmap pattern. It defines the consumer-facing OpenShift cluster service separately from the platform architecture narrative and the ABB/SBB catalog.

The offering is intentionally versioned. CLaaS 0.1.0 is the first modeled increment: requestable HCP-based OpenShift clusters with virtualized workers and explicit launch blockers.

Service Definition

CLaaS provides governed OpenShift cluster capacity for business teams and platform services. The DC 3.0 target state is a self-service, policy-governed, evidence-producing cluster service with standard profiles, automated lifecycle, integrated IAM, controlled endpoint publication, observability, backup, and clear support boundaries.

The offering definition answers what a consumer receives. Architecture notes and SBB notes explain how the platform provides it.

Roadmap Model

The machine-readable model lives in sidecar YAML next to this offering note:


RWS-ARCH/70-togaf/offerings/clusters-as-a-service-offering.model.yaml

That sidecar is the entry point for the CLaaS dependency chain. The portal and validators include it automatically by scanning RWS-ARCH/**/*.model.yaml.

Dependencies

CLaaS 0.1.0 currently depends on separate capability sidecars for:

  • fleet HCP cluster provisioning
  • cluster identity access
  • cluster ingress publication
  • cluster baseline GitOps
  • cluster observability baseline
  • cluster backup and restore baseline

The model derives offering availability from the latest required capability window, not from a separately maintained manual launch date.

Open Questions

  • Which CLaaS 0.1.0 dependencies can be scoped out for non-production-only consumers?
  • Should backup and restore evidence block initial availability or only production onboarding?
  • Which exact service profiles belong in CLaaS 0.1.0 versus later increments?
  • Which team owns the update cadence for each capability state and target window?