Radio Ecosystem Orchestrator (Radio Infra)

Orchestration, observability, and deployment repository for the Radio ecosystem microservices.

This repository does not contain application code — it wires together the independent radio-registry and radio-analytics microservices via Docker Compose, and drives their automated deployment to Amazon Web Services (AWS).

Microservices application traces, metrics, and logs are collected by Alloy and stored in Prometheus (metrics), Tempo (distributed traces), and Loki (logs). Grafana ties the three signals together for visualization, dashboards, and trace-to-log correlation.

🚧 Work in Progress

This project is currently under active development and serves as a demonstration of modern DevOps and cloud infrastructure practices.

Live Demo

Grafana dashboards are available at: radio-analytics.giuliopetteno.dev (short link for click tracking)

• Radio Analytics - Insights → Business analytics for medical devices, including inventory, allocation, lifecycle, organizational structure, and activity trends, available both in real time and as scheduled, persisted reports
• Radio Analytics - Performance → Analytics engine execution monitoring, including outcomes, execution times, KPI status and performance, across both real-time and scheduled report executions
• Radio Ecosystem - Operations → End-to-end observability of the entire ecosystem, including service dependencies, HTTP and Kafka traffic, logs, and distributed traces

Note: Anonymous read-only access — no login required.

Features

• Containerization
• Automated CI/CD pipeline
• Cloud deployment
• Full observability stack: distributed tracing, metrics, and structured logging with cross-signal correlation

Architecture

flowchart LR
  Client([Client])

  subgraph aws["AWS"]
    subgraph ec2["AWS EC2"]
      Nginx[nginx]

      subgraph compose["Docker Compose · radio-infra"]
        subgraph network["Docker network · radio-net"]
          Registry[radio-registry]
          Analytics[radio-analytics]
          Kafka[Apache Kafka]
          Alloy[Alloy]
          Prometheus[Prometheus]
          Tempo[Tempo]
          Loki[Loki]
          Grafana[Grafana]
        end
      end
    end

    RDS["AWS RDS · PostgreSQL"]
  end

  Client --> Nginx
  Nginx --> Registry
  Nginx --> Grafana

  Registry -- radio_registry schema --> RDS
  Analytics -- radio_analytics schema --> RDS

  Registry --> Kafka
  Kafka --> Analytics

  Grafana --> Analytics

  Registry -. OTLP .-> Alloy
  Analytics -. OTLP .-> Alloy
  Alloy -. metrics .-> Prometheus
  Alloy -. logs .-> Loki
  Alloy -. traces .-> Tempo

  Grafana --> Prometheus
  Grafana --> Loki
  Grafana --> Tempo

  style aws fill:#fff,stroke:#d0d7de
  style ec2 fill:#fff,stroke:#ed7100,stroke-width:2px
  style compose fill:#f6f8fa,stroke:#d0d7de
  style network fill:#fff,stroke:#d0d7de,stroke-dasharray:5 5

  classDef app fill:#f6f8fa,stroke:#d0d7de,color:#24292f
  classDef proxy fill:#e8467c,stroke:#e8467c,color:#0d1117
  classDef broker fill:#79c0ff,stroke:#79c0ff,color:#0d1117
  classDef metrics fill:#e6522c,stroke:#e6522c,color:#fff
  classDef grafana fill:#f9a825,stroke:#f9a825,color:#0d1117
  classDef awsService fill:#ed7100,stroke:#ed7100,color:#fff

  class Registry,Analytics,Client app
  class Nginx proxy
  class Kafka broker
  class Prometheus metrics
  class Alloy,Tempo,Loki,Grafana grafana
  class RDS awsService

Loading

• radio-registry — producer service (medical devices management system)
• radio-analytics — consumer service (medical devices analytics system)
• Services communicate exclusively via Apache Kafka; they share no code and are versioned in separate repositories
• A shared Docker network (radio-net) connects all containers
• Application code lives in the service repositories; the message broker and observability stack are defined and orchestrated here

Deployment Modes

This repo supports three distinct Compose configurations:

• docker-compose.yml — local development. Uses Compose include to pull in each service's own compose file from sibling directories, building images from local source
• docker-compose.prod.yml — production-like local environment. Uses Compose include to pull in each service's own .prod compose file from sibling directories, building images from local source
• docker-compose.aws.yml — cloud production. Self-contained (no dependency on sibling repos being cloned), pulls pre-built images directly from Amazon ECR, and is the configuration deployed to EC2

CI/CD Pipeline

flowchart LR
  subgraph github["GitHub"]
    subgraph repos["Repositories"]
      RegistryRepo[radio-registry]
      AnalyticsRepo[radio-analytics]
      InfraRepo[radio-infra]
    end

    subgraph ci["CI · GitHub Actions"]
      Build[Build and test]
      Image[Build container image]
    end

    subgraph cd["CD · GitHub Actions"]
      Deploy[SSM Run Command]
    end
  end

  subgraph aws["AWS"]
    ECR[AWS ECR]

    subgraph ec2["AWS EC2"]
      Compose[Docker Compose<br/>radio-infra]
    end
  end

  RegistryRepo -- push --> Build
  AnalyticsRepo -- push --> Build

  Build --> Image
  Image -- docker push · OIDC --> ECR
  Image -- repository_dispatch --> InfraRepo

  InfraRepo --> Deploy
  Deploy -- OIDC --> Compose
  Compose -- docker pull --> ECR

  style github fill:#fff,stroke:#d0d7de
  style repos fill:#fff,stroke:#d0d7de
  style ci fill:#fff,stroke:#d0d7de
  style cd fill:#fff,stroke:#d0d7de
  style aws fill:#fff,stroke:#d0d7de
  style ec2 fill:#fff,stroke:#d0d7de

  classDef repo fill:#f6f8fa,stroke:#d0d7de,color:#24292f
  classDef pipeline fill:#a371f7,stroke:#a371f7,color:#0d1117
  classDef awsService fill:#ed7100,stroke:#ed7100,color:#fff

  class RegistryRepo,AnalyticsRepo,InfraRepo repo
  class Build,Image,Deploy pipeline
  class ECR,Compose awsService

Loading

Technology Stack

• Containerization with Docker and Docker Compose
• Automated CI/CD with GitHub Actions
• Amazon Web Services (AWS) deployment:
  • EC2 (Docker Compose orchestration, IAM-only access via SSM)
  • ECR for container image registry
  • RDS (PostgreSQL, private subnet, EC2-scoped security group, SSM tunnel for local dev)
  • GitHub Actions → OIDC → ECR → SSM Run Command deploy
  • IAM: least-privilege roles throughout (GitHub Actions OIDC roles, EC2 instance role)
  • Secrets management via AWS Systems Manager Parameter Store
  • Nginx reverse proxy for name-based routing, with TLS via Let's Encrypt and automated renewal
  • DNS-based service routing under a custom domain (giuliopetteno.dev) via Route 53 with Elastic IP
• Observability stack:
  • Alloy as unified OpenTelemetry (OTLP) collector
  • Prometheus for metrics storage
  • Tempo for distributed trace storage
  • Loki for log aggregation
  • Grafana for dashboards, visualization, and trace-to-log correlation