As network engineers and DevOps practitioners, we frequently need to map out both logical connections and physical infrastructure layouts. While commercial tools like Visio or Lucidchart dominate the space, their licensing costs create barriers for individual developers and small teams.

After testing multiple solutions, these three options stand out for their capabilities:

1. Draw.io (now Diagrams.net)
   - Browser-based with offline mode
   - Extensive icon libraries for Cisco, AWS, Azure
   - Export formats: PNG, SVG, XML (for version control)

2. yEd Graph Editor
   - Java-based desktop application
   - Advanced auto-layout algorithms
   - Supports BPMN and UML alongside network diagrams

3. PlantUML
   - Text-to-diagram approach (perfect for coders)
   - Integrates with Markdown/docs-as-code workflows
   - Sample network definition:
     @startuml
     node "Web Server" as ws
     node "Database" as db
     ws --> db : JDBC
     @enduml

For physical topology mapping, NetBox offers an interesting open-source approach:

# Sample YAML for rack definition in NetBox
racks:
  - name: Rack-01
    facility_id: DC-01-A12
    devices:
      - name: core-switch-01
        role: core-switch
        position: 42
        manufacturer: Cisco

Logical diagrams benefit from Graphviz, especially when generating from configuration:

// Sample DOT language for Graphviz
digraph G {
  subgraph cluster_0 {
    label="AWS VPC";
    public_subnet -> internet_gateway;
    private_subnet -> nat_gateway;
  }
  public_subnet -> load_balancer -> app_tier;
}

The real power emerges when combining these tools with your existing pipelines:

# Terraform + Graphviz example
terraform graph | dot -Tsvg > infrastructure.svg

# Ansible + Diagram generation
- name: Generate network diagram
  community.general.diagrams:
    source: "{{ playbook_dir }}/network.yml"
    output: "/var/www/network.png"

For teams practicing GitOps, storing diagram sources alongside infrastructure code ensures documentation never drifts from reality.

As a developer who frequently needs to document network architectures, I've faced the same frustration as many others - most professional-grade network diagram tools like Microsoft Visio or Lucidchart require expensive subscriptions. After extensive testing of alternatives, here's my curated list of truly free solutions that support both logical and physical network mapping.

// Pseudocode for evaluating diagram tools
const requirements = {
  isFree: true,
  openSource: preferred,
  logicalDiagram: true, 
  physicalDiagram: true,
  exportFormats: ['PNG','SVG','PDF'],
  collaboration: optional
};

const testResults = [
  diagramTools.filter(tool => 
    meetsRequirements(tool, requirements)
  )
];

My personal favorite for quick prototyping. The web version requires no installation:

// Sample network diagram XML (simplified)

  
    
    
    
  

Key advantages include real-time collaboration and VSCode integration via extensions.

Java-based offline tool with superior auto-layout algorithms. Sample workflow:

1. Import node list from CSV:
   Device,IP,Role
   R1,192.168.1.1,Core
   SW1,192.168.1.2,Access

2. Apply "Organic Layout" 
3. Export as interactive HTML

Perfect for developers who prefer defining networks as code:

// Physical topology in Diagram.Codes syntax
network PhysicalDC {
  rack Rack1 {
    device: Router [model: CiscoASR]
    device: Switch [stack: 2] 
    device: Server [role: Database]
  }
  
  connections:
    Router -> Switch [cables: fiber]
    Switch -> Server [ports: eth1-4]
}

For programmatic generation from live network data:

import graphviz

def generate_topology(devices):
    dot = graphviz.Digraph(comment='Network Map')
    for device in devices:
        dot.node(device.hostname, shape='rectangle')
    for link in device.connections:
        dot.edge(link.source, link.target)
    return dot

# Usage: 
print(generate_topology(scan_network('192.168.1.0/24')))

For sensitive environments, these open-source tools can be deployed internally:

• NetBox (Django-based with diagram plugins)
• OpenDCIM (Data center focused)
• phpIPAM (IP-focused visualization)

For most developers, Diagrams.net provides the best balance of features and accessibility. However, if you're working with dynamic networks that change frequently, the Python+Graphviz approach offers maximum flexibility.