Chapter 06: Orchestration (The Captain)

"It works on my machine" is not a valid excuse.

This optional deployment sketch discusses five components; the repository does not include a working GoTracker or Kafka deployment:

1. Gopher Shop (Web + API)
2. Postgres (DB)
3. GoTracker (Microservice)
4. Redis (Cache)
5. Kafka (Broker)

Starting them manually (5 terminal windows) is madness.

1. Docker Compose (The Local Conductor)

We define the whole orchestra in one file: docker-compose.yml.

The following is a structural sketch, not a runnable Compose configuration. It omits database credentials and volumes, Kafka configuration, health checks, and the application Dockerfiles. Do not treat it as an installation recipe.

services:
  shop:
    build: ./gopher-shop
    ports: ["8080:8080"]
    depends_on: [postgres, kafka]
  
  tracker:
    build: ./go-tracker
    depends_on: [redis, kafka]

  postgres:
    image: postgres:15
  
  redis:
    image: redis:7
    
  kafka:
    image: bitnami/kafka:latest

For an actual validated configuration, the command is docker compose up. A short depends_on orders startup but does not establish service readiness. Add health checks and application retry behavior; pin and test image versions before deployment.

2. Kubernetes (The Container Ship) 🚢

Kubernetes is an optional orchestration platform. A small service can run on a single host or a managed application platform; production does not require a cluster. Choose it only when scheduling and operational needs justify its cost.

• Pod: A small group of containers (e.g., 1 Shop instance).
• Deployment: "I want 3 replicas of the Shop."
• Service: "Give me a Load Balancer IP to talk to the Shop."

3. Visual Signal (The Container Ship) ⚓

Concept: Container Orchestration. Signal: A Captain managing a massive ship.

• Containers: The standard metal boxes.
• Docker: The crane that lifts them.
• Kubernetes: The Captain who decides where each box goes to balance the ship.

4. Final Words on Ops

As a Backend Engineer, you don't need to be a DevOps expert. But you MUST understand how your code runs.

• How do I view logs? (kubectl logs)
• How do I restart it? (kubectl rollout restart)
• How do I scale it? (kubectl scale --replicas=5)