Robotron: Top-down network management at scale

This post introduces Robotron, a top-down network management system designed to manage tens of thousands of network devices and hundreds of thousands of servers globally at Facebook. Robotron aims to reduce manual interaction with network devices for tasks like circuit turn-up, migration, device provisioning, OS upgrades, ACL modification, protocol tuning, and monitoring. It evolved from an earlier system, FBNet, which started as an object store for basic network device information and evolved to store normalized, vendor-agnostic views of network state. Robotron covers network design, configuration generation (vendor-specific), deployment (phased rollouts), and monitoring. The system translates high-level engineer intent into FBNet objects, then generates vendor-specific configurations, deploys them safely, and monitors for deviations from the desired state. The post highlights the challenges of distributed configurations, multiple domains, versioning, dependencies, and vendor differences, and mentions the open-sourcing of parts of Robotron like fbpush.