Docker Network Isolation and Inter-Container Security

Overview

By default, Docker containers on the default bridge network can communicate with each other. This is a security risk: a compromised container can probe or attack other containers on the same host. The pattern below demonstrates how to isolate containers using custom networks, internal networks, and access control to limit which containers can talk to each other.

When to Use

• You run multiple services on the same host and need to restrict inter-container communication
• You have a public-facing container (API) and a private container (database) that should not be reachable from outside
• You want to segment services by trust level (frontend, backend, database)
• You need to comply with security policies requiring network segmentation

Solution

Default bridge vs custom bridge

# Default bridge: containers can talk to each other (insecure)
docker run -d --name api --network bridge my-api
docker run -d --name db --network bridge my-db
# api can reach db and vice versa — no isolation

# Custom bridge: containers can only talk to containers on the same network
docker network create --driver bridge frontend-net
docker network create --driver bridge backend-net

docker run -d --name api --network backend-net my-api
docker run -d --name db --network backend-net my-db
docker run -d --name web --network frontend-net my-web

# web cannot reach db — different network
# api can reach db — same network

Internal network (no internet access)

# Create an internal network — containers cannot reach the internet
docker network create --driver bridge --internal backend-internal

docker run -d --name db --network backend-internal my-db
# db has no internet access, only inter-container communication on this network

Multi-network container (API connects to both frontend and backend)

docker network create frontend-net
docker network create backend-net

# API container joins both networks
docker run -d --name api --network frontend-net my-api
docker network connect backend-net api

docker run -d --name web --network frontend-net my-web
docker run -d --name db --network backend-net my-db

# web -> api (frontend-net) ✓
# api -> db (backend-net) ✓
# web -> db ✗ (different networks)

Docker Compose with network isolation

# docker-compose.yml
services:
    web:
        image: nginx:alpine
        ports:
            - "80:80"
        networks:
            - frontend
        depends_on:
            - api

    api:
        build: .
        networks:
            - frontend
            - backend
        depends_on:
            db:
                condition: service_healthy

    db:
        image: postgres:16-alpine
        networks:
            - backend
        environment:
            POSTGRES_PASSWORD: ${DB_PASSWORD}
        healthcheck:
            test: ["CMD", "pg_isready", "-U", "postgres"]
            interval: 10s
            timeout: 5s
            retries: 5

networks:
    frontend:
        driver: bridge
    backend:
        driver: bridge
        internal: true   # No internet access for backend

Network with IP range and subnet

docker network create \
    --driver bridge \
    --subnet 172.20.0.0/16 \
    --ip-range 172.20.0.0/24 \
    --gateway 172.20.0.1 \
    backend-net

Restricting published ports with IP binding

# Bind to localhost only — not accessible from other machines
docker run -d -p 127.0.0.1:5432:5432 --name db postgres:16-alpine

# Bind to a specific interface
docker run -d -p 10.0.0.5:80:80 --name web nginx:alpine

# Bind to all interfaces (default — least secure)
docker run -d -p 0.0.0.0:80:80 --name web nginx:alpine

Overlay network for Docker Swarm

# Create an overlay network (requires Swarm mode)
docker network create --driver overlay --attachable my-overlay-net

# Services on different hosts can communicate securely
docker service create --name api --network my-overlay-net my-api
docker service create --name db --network my-overlay-net my-db

Inspecting network connectivity

# List all networks
docker network ls

# Inspect a network to see connected containers
docker network inspect backend-net

# Test connectivity from one container to another
docker exec api ping db
docker exec api curl -f http://db:5432

# Remove a container from a network
docker network disconnect backend-net api

Explanation

Docker networks provide isolation at the data link layer. Containers on different networks cannot communicate directly.

Network types:

• bridge: The default driver for single-host networking. Creates a virtual ethernet bridge on the host. Custom bridge networks provide DNS resolution (container names resolve to IPs) and isolation.
• internal: A bridge network with --internal flag. Containers on this network have no external internet access. Useful for databases that should only be reachable by application containers.
• overlay: For multi-host networking with Docker Swarm. Uses VXLAN tunnels between hosts. Containers on the same overlay network can communicate across hosts.
• host: No isolation — the container uses the host’s network stack directly. Use only for performance-critical scenarios.

Security principles:

• Least privilege: Each container should only be on the networks it needs. A database does not need internet access.
• Segmentation: Separate public-facing services (web) from internal services (database) using different networks.
• Internal networks: Use internal: true for networks that should have no internet access.
• Port binding: Bind to 127.0.0.1 for services that should only be accessible from the host, not from the network.

Variants

Network Type	Scope	Internet Access	Use When
Default bridge	Single host	Yes	Development only
Custom bridge	Single host	Yes	Production single-host
Internal bridge	Single host	No	Databases, private services
Overlay	Multi-host (Swarm)	Yes	Swarm clusters
Host	Single host	Yes	Maximum performance

Guidelines

• Never use the default bridge network in production. Create custom networks.
• Use internal: true for backend networks containing databases and private services.
• Connect containers to only the networks they need. A web container should not be on the backend network.
• Bind database ports to 127.0.0.1 only. Never expose databases to external interfaces.
• Use Docker Compose networks to define segmentation declaratively.
• Use overlay networks for Swarm multi-host communication.
• Inspect networks regularly with docker network inspect to verify connectivity.
• Remove unused networks with docker network prune.

Common Mistakes

• Using the default bridge network in production. All containers can reach each other.
• Exposing database ports to 0.0.0.0. Anyone on the network can connect.
• Putting all containers on one network. No segmentation means a compromised container can attack everything.
• Not using internal: true for backend networks. Databases can make outbound internet connections.
• Forgetting that containers on different networks cannot resolve each other’s names. DNS only works within the same network.
• Connecting a container to too many networks. This increases the attack surface.

Performance Tips

1. Use host network for maximum throughput. Bypasses the Docker bridge entirely:

# No NAT overhead, but no isolation
docker run -d --network host my-app

Only use this for performance-critical workloads where isolation is handled at a different layer.

2. Reduce DNS lookup overhead. Docker’s embedded DNS adds latency for each lookup:

# Add DNS cache in container
docker run -d --dns 127.0.0.11 --dns-opt "timeout:1" --dns-opt "attempts:1" my-app

3. Use --network-alias for faster service discovery. Avoids container name lookups:

services:
  db:
    networks:
      backend:
        aliases:
          - db.internal