Deploy IoT Applications with Docker and Docker Compose

Why Docker for IoT?

In traditional IoT projects, installing software on hundreds of edge devices is nightmare — each device has different OS version, libraries, and configuration. Docker solves this by packaging entire application with dependencies in container, ensuring runs identically everywhere.

IoT Architecture with Docker

Typical IoT system has multiple services on edge device:

┌─────────────────────────────────┐
│         Edge Device (ARM64)      │
│  ┌───────────┐ ┌──────────────┐ │
│  │ MQTT      │ │ Data         │ │
│  │ Broker    │ │ Collector    │ │
│  └───────────┘ └──────────────┘ │
│  ┌───────────┐ ┌──────────────┐ │
│  │ InfluxDB  │ │ Grafana      │ │
│  │ (TimeSeries) │ (Dashboard) │ │
│  └───────────┘ └──────────────┘ │
└─────────────────────────────────┘

Docker Compose for Multi-Service IoT

docker-compose.yml manages entire stack:

version: "3.8"
services:
  mosquitto:
    image: eclipse-mosquitto:2.0
    ports:
      - "1883:1883"
    volumes:
      - ./mosquitto/config:/mosquitto/config
      - mosquitto_data:/mosquitto/data
    restart: unless-stopped

  collector:
    build: ./collector
    environment:
      - MQTT_HOST=mosquitto
      - INFLUX_HOST=influxdb
    depends_on:
      - mosquitto
      - influxdb
    restart: unless-stopped

  influxdb:
    image: influxdb:2.7-alpine
    volumes:
      - influx_data:/var/lib/influxdb2
    environment:
      - DOCKER_INFLUXDB_INIT_MODE=setup
      - DOCKER_INFLUXDB_INIT_USERNAME=admin
    restart: unless-stopped

  grafana:
    image: grafana/grafana:10.2-alpine
    ports:
      - "3000:3000"
    volumes:
      - grafana_data:/var/lib/grafana
    restart: unless-stopped

volumes:
  mosquitto_data:
  influx_data:
  grafana_data:

Optimize Docker for Edge Devices

1. Multi-stage Build Reduces Image Size

# Build stage
FROM python:3.11-slim AS builder
WORKDIR /app
COPY requirements.txt .
RUN pip install --no-cache-dir --prefix=/install -r requirements.txt

# Runtime stage
FROM python:3.11-slim
COPY --from=builder /install /usr/local
COPY . /app
WORKDIR /app
CMD ["python", "collector.py"]

Image reduces from ~900MB to ~150MB — critical when deploying over 4G network.

2. Multi-arch Build for ARM64

Many edge devices (Raspberry Pi, NVIDIA Jetson) run ARM64:

# Create buildx builder
docker buildx create --name iotbuilder --use

# Build for both AMD64 and ARM64
docker buildx build --platform linux/amd64,linux/arm64 \
  -t myregistry/collector:v1.2 --push .

3. Resource Limits

Edge devices have limited RAM, always set limits:

services:
  collector:
    deploy:
      resources:
        limits:
          memory: 256M
          cpus: "0.5"

CI/CD for IoT Fleet

Automatic Deployment Workflow

# 1. Push code to GitHub
git push origin main

# 2. GitHub Actions builds and pushes image
# 3. On each device, Watchtower auto-updates

Use Watchtower for automatic image pulling:

services:
  watchtower:
    image: containrrr/watchtower
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock
    environment:
      - WATCHTOWER_POLL_INTERVAL=300
      - WATCHTOWER_CLEANUP=true
    restart: unless-stopped

Container Security for IoT

Never run as root: use user: "1000:1000" in compose
Read-only filesystem: read_only: true for containers that don't write
Limit capabilities: security_opt: [no-new-privileges:true]
Secrets management: Docker secrets or .env file with chmod 600

For larger IoT fleet management, see Kubernetes for Robot Fleet with K3s and GitOps.

Monitoring and Logging

On edge devices, log rotation critical to avoid filling disk:

services:
  collector:
    logging:
      driver: "json-file"
      options:
        max-size: "10m"
        max-file: "3"

Docker and Docker Compose let you manage IoT applications professionally — from development to production, from single device to hundreds. Combined with CI/CD and auto-updates, deploy and maintain IoT fleet without SSH to each device.