How o-ran-sc/nonrtric works
Project Type and Purpose
• This is a microservices‐based, containerized orchestration platform (Non-RT RIC) for O-RAN, implemented in Go, Java and Python. It provides non‐real‐time RAN policy and AI/ML model management via A1 interfaces, orchestrates message flows over DMaaP (Kafka), and exposes services via Kubernetes/Helm/Istio.
File Structure
• Top‐level directories:
• .github, .releases, CI/CD and release manifests (Helm charts & container/K8s YAML).
• auth-token-fetch: a Go microservice fetching auth tokens.
• docker-compose: local dev/test environments for each microservice (A1 sim, DMaaP sim, policy framework, ICS, RAPP, SDNC, etc.).
• docs: Sphinx documentation and design guidance.
• sample-services: stand-alone example R-APPs (hello-world, ICS producer/consumer) with Docker, Helm charts.
• service-exposure: Go CLI and K8s/Istio/Keycloak/Cert-Manager scripts and Helm charts for exposing services.
• test: automated shell/python test harnesses, simulators, certificate generators.
• Key patterns:
• Microservices communicate via Kafka (DMaaP) topics.
• A1 Controller and policy management services expose REST/gRPC A1 interfaces.
• A policy-framework uses Drools/XACML PDPs.
• Helm charts for Kubernetes packaging.
• Istio Gateway + MutatingWebhookConfiguration for secure ingress with Keycloak and JWT.
README for Additional Insights
• Confirms focus on non-real-time RIC functions: policy optimization, AI/ML model distribution, coordination with Near-RT RIC.
• No monolithic UI—pure back-end services.
a. Main Components
• Non-RT RIC Microservices:
• A1-Controller
• A1-Policy-Management-Service
• Policy-Agent
• Enrichment Coordinator
• Information Coordinator
• R-APP Catalogue Controller
• O-DU Slice Assurance, O-RU Closed-Loop Consumer/Recovery
• Auth-Token-Fetch Service
• Message Bus: DMaaP (Kafka) topics for events in/out.
• Policy Framework: Drools PDP, XACML PDP, database (embedded or external).
• External Systems:
• Near-RT RIC (A1 interface).
• SDNC, O-Cloud, O-CAPIF.
• Keycloak for auth/OIDC.
• Service Exposure Layer:
• Istio Gateway, VirtualService, EnvoyFilter.
• Kubernetes Ingress + JWT Authentication Webhook.
• Sample R-APPs: example producer/consumer apps.
b. Relationships and Interactions
• Near-RT RIC ←→ A1-Controller (REST/gRPC).
• A1-Controller → A1-Policy-Management → Policy-Agent → DMaaP (publish policies).
• Enrichment Coordinator subscribes DMaaP → processes → publishes enriched data.
• Information Coordinator collects jobs from DMaaP → stores/forwards to R-APPs.
• R-APP Catalogue Controller manages R-APP registrations → policy updates.
• Auth-Token-Fetch provides JWTs to microservices for secure calls.
• Service Exposure Layer front-ends A1 and sample-service APIs to external clients, using Istio + Keycloak.
• Policy Framework PDPs invoked by Policy-Agent via HTTP or gRPC.
c. Architectural Patterns & Principles
• Microservices communicating asynchronously via pub/sub.
• API Gateway (Istio) and sidecar proxies enforcing security policies.
• Helm chart packaging for K8s deployment and Rollback.
• Separation of concerns: ingestion, coordination, policy management, exposure.
d. Relevant Technologies
• Go, Java Spring Boot, Python (test harness), Kafka/DMaaP, Docker, Helm, Kubernetes, Istio, Keycloak, Drools/XACML PDP, GitHub Actions / Gerrit CI.