NVIDIA Config Manager is a Helm-deployed network automation and configuration management platform designed to bootstrap and operate NVIDIA AI-factory networks. It combines APIs, workers, Nautobot, data stores, eventing, and device-facing services to render, deploy, and operate network configuration at scale.
Overview
Running on Nautobot, Config Manager provides a comprehensive suite of services for network automation:
Design Principles
Config Manager follows these operating principles:
Platform Independence
- Does not rely on a single network operating system (NOS) technology suite
- Supports multiple vendor platforms and OS versions
- Design and configuration agnostic architecture
Simplified Operations
- Eliminates intermediate orchestration layers (such as Ansible or SaltStack)
- Uses direct workflow-based automation for bootstrap and change operations
- Gives network engineers one workflow path for provisioning, validation, and deployment
Self-Contained System
- Fully self-contained with no external dependencies
- All components are either open source or developed in-house
- Complete control over the technology stack
Zero-Touch Bootstrap
- Bootstraps networks without console or CLI access to each device
- Automates provisioning from device boot through intended configuration
- Supports repeatable bring-up for new sites and replacement devices
Configuration Agnostic
- No embedded logic in the system
- Design and configuration managed externally
- Flexible architecture supports any network design
SOC2 Compliance
- Incorporates approval workflows for Day Two changes
- Audit trail for all configuration changes
- Change management and governance built in
Live Data Validation
- Uses live data sources to validate topology
- Validates cabling against intended design
- Detects and reports discrepancies
Version Controlled
- All code and configuration under source control
- Full version history and rollback capability
- GitOps-based configuration management