GNS3 Documentation
website·Indexed 36 days ago
https://docs.gns3.com/Documentation for GNS3, a network simulation tool used for experimentation and learning. It covers the GNS3 VM, server administration, and the use of emulators like QEMU, Docker, and Cisco IOU. The guides include details on network topology configuration, NAT nodes, SSL encryption for client-server communication, and appliance template management for versions 2.0 and 3.1.
What's inside GNS3
- The GNS3 Server provides a Model Context Protocol (MCP) interface that allows AI assistants (such as Claude Code) to interact with GNS3 using the Server-Sent Events (SSE) transport protocol. This enables users to manage GNS3 projects, nodes, links, templates, and compute nodes using natural language commands instead of the GUI/Web UI. It can also be used to build custom AI Agents or integrate GNS3 into automated testing workflows.
Overview of GNS3 MCP Service
GNS3 Server implements the Model Context Protocol (MCP), an open standard that allows AI assistants (such as Claude Code) to interact with GNS3 via natural language. This enables the management of projects, nodes, links, templates, and compute nodes without using the GNS3 GUI or Web UI. The service uses SSE (Server-Sent Events) transport.Overview of Virtual PC Simulator (VPCS) in GNS3
VPCS (Virtual PC Simulator) is a lightweight tool in GNS3 used to simulate a minimal PC environment for testing network connectivity. Unlike full virtual machines (VMware/QEMU), VPCS consumes very few resources (< 5MB RAM), boots in seconds, and provides only basic IP and networking tools rather than a full operating system. It is ideal for connectivity tests and protocol experiments.Overview of Equinix Metal for GNS3
Equinix Metal is a bare metal cloud provider that offers dedicated servers rather than virtual machines. This provides direct access to CPU virtualization instructions, making it a highly performant environment for running GNS3 on Packet.Overview of GNS3 3.1 New Features
GNS3 3.1 introduces several major updates to the platform:
- New Web UI: Now supports Web Console, Web VNC, and Web Wireshark directly in the browser.
- Frontend Upgrade: The Web UI framework has been upgraded from Angular 14 to Angular 21.
- AI Integration: New AI-driven tools including an AI Assistant, MCP service for external AI clients, AI Skills, fault injection for testing, and real-time packet analysis.
- GNS3 VM: Improved performance and usability.
- Multi-user Collaboration: Enhanced support for teamwork and permission management.
Overview of GNS3-Skills knowledge base
GNS3-Skills is the domain knowledge base for GNS3 Copilot, providing a knowledge layer for GNS3 network simulation environments via LangChain tools. It enables automated fault diagnosis, experiment automation, and learning assistance. The library covers 60+ network protocols and technologies (ranging from PPP to SRv6 and EVPN), containing 50 YAML skill files and 815 fault injection scenarios.Understand GNS3 Architecture and Server Options
GNS3 consists of two primary software components: the GNS3-all-in-one software (GUI client) and the GNS3 virtual machine (VM). The GUI is installed on the local PC (Windows, Mac, Linux) to create topologies, but the devices within those topologies must be hosted by a server process.
There are three server hosting options:
- Local GNS3 server: Runs as a process on the same PC as the GUI. This is the simplest setup but is limited in terms of topology size and supported devices.
- Local GNS3 VM: Runs locally via virtualization software (VMware Workstation, VirtualBox, or Hyper-V). This is recommended for advanced topologies and devices requiring Qemu (e.g., Cisco VIRL devices).
- Remote GNS3 VM: Runs on a remote server using VMware ESXi or in the cloud.
Compare Dynamips IOS images vs Routing and Switching VMs
GNS3 provides two primary ways to emulate Cisco IOS: Dynamips and Routing and Switching virtual machines (R+Svms).
- Dynamips: Used for older, unmodified IOS images. They have lower CPU and memory overhead, allowing for more devices in a single topology.
- R+Svms: Vendor images housed in virtual machines. They are larger in size but provide more realistic real-world network emulation.
Use the GNS3 Ethernet Hub for Layer 1 simulation
The Ethernet Hub is a built-in GNS3 node that simulates a simple physical layer (Layer 1) repeater. It is primarily used for educational purposes to demonstrate shared bandwidth, collision domains, and CSMA/CD concepts.
Key Characteristics:
- Behavior: Pure physical layer broadcasting; data received on one port is forwarded to all other ports.
- Ports: Defaults to 8 ports (Ethernet0 to Ethernet7).
- Management: Built-in nodes are always running upon creation; they do not support start/stop operations or console access.
- Monitoring: Supports PCAP packet capturing on individual ports.
Use the GNS3 Ethernet Hub for Layer 1 simulation
The Ethernet Hub is a built-in GNS3 node that simulates a physical layer (Layer 1) repeater. It is primarily used for teaching concepts like collision domains and CSMA/CD. It uses the Dynamipsnio_bridgehypervisor command. When the Hub receives data, it broadcasts that data to all ports except the incoming port. It supports PCAP packet capture on individual ports.Simulate network problems using AI Fault Injection
The Fault Injection feature combines AI with real device settings in a topology to simulate network problems. It executes predefined faults sourced from thegns3-skillsrepository by running specific configuration commands on devices. This is primarily used for practicing fault recovery and learning protocol behavior.GNS3 Use Cases and Capabilities
GNS3 is used for learning, teaching, and professional network engineering. Key capabilities include:
- Certification Prep: Practicing for exams like Cisco CCNA.
- Proof of Concept (POC): Testing multi-vendor interoperability and SDN software without physical hardware.
- Pre-deployment Testing: Running OSs that emulate real hardware behavior to test networks before building them.
- Physical Integration: Connecting virtual GNS3 topologies directly to real physical networks.