How k0rdent/kof works
Project type and purpose
• This is a Kubernetes-native Observability and FinOps platform (“kof”) that comprises:
• A Go-based Kubernetes Operator (“kof-operator”) that uses CRDs to coordinate multi-cluster deployments and telemetry aggregation
• A suite of Helm charts under charts/ to install components like collectors (OpenTelemetry, Prometheus, VictoriaMetrics), Istio configuration, storage, and dashboards
• Demo manifests and extensive documentation for running on AWS, Azure, EKS, etc.
• Its goal is to deploy, configure, and manage telemetry in child and regional clusters, aggregate metrics/traces in a “mothership,” and present dashboards & alerts for cost-and-performance insights.
File structure
Top-level directories reveal architectural layers:
• charts/ – Helm charts for each logical component: collectors, Istio config, “mothership” aggregator, storage backends, operators, child/regional multi-cluster services
• kof-operator/ – the custom Kubernetes Operator built with controller-runtime, defining CRDs (PromxyServerGroup, MultiClusterService, ClusterDeployment, etc.), reconcilers, RBAC, and Dockerfiles
• demo/ – example values and cluster manifests showing how operator and charts are applied in real environments
README
• Confirms project focus (“k0rdent Observability and FinOps”) and points at full documentation
• Suggests conventional commit standards for contribution
Final advice
• CRDs (e.g., PromxyServerGroup, MultiClusterService) as diamonds or annotations to underscore the operator pattern
• Keep the layout clean: control flows on top, data flows bottom, storage at the center or right side
• With version or Helm values file names where meaningful (e.g., demo/demo-mothership-values.yaml)