# Baseboard Management Controller Inside every server is a second, much smaller computer with its own processor, memory, firmware, and network connection, powered whenever the machine has power at all. It watches the hardware, reports its condition, and can turn the server on, off, and back on again whether or not the operating system is alive. Without it, a fleet of a hundred thousand machines would require a hundred thousand visits. > [!note] The one-sentence version > A little computer whose whole job is to watch and control the big computer, including when the big computer is dead. ## Why it exists Servers fail in ways that remove exactly the tools needed to diagnose them. An operating system that will not boot cannot report why; a machine wedged in firmware answers no network request; a node with a failed memory module presents nothing to log into. Management hardware solves this by being independent: separate silicon, separate firmware, separate power domain, separate network. From the operator's console a server that is entirely unresponsive can still be interrogated, power-cycled, re-imaged, and watched through its boot sequence. ## What it does - **Power control.** On, off, cycle, and hard reset, independent of the operating system. - **Console redirection.** Presenting the machine's video, keyboard, and mouse over the network, so an engineer can watch firmware screens and boot messages remotely. - **Virtual media.** Attaching an installation image across the network, which is how bare machines are provisioned and recovered; see [[wiki/Bare Metal Compute|Bare Metal Compute]]. - **Sensor telemetry.** Temperatures, voltages, currents, fan speeds, power draw, and component health, streamed into monitoring; this is the data behind power and thermal capacity management; see [[wiki/Observability|Observability]]. - **Event logging.** A persistent record of hardware events that survives the operating system's failure. - **Firmware management.** Updating the system firmware, the controller's own firmware, and device firmware across a fleet. - **Inventory and attestation.** Reporting precisely what hardware is present and, increasingly, cryptographically attesting that firmware is what it should be; see [[wiki/Supply Chain Security|Supply Chain Security]]. Modern controllers are driven through a standardized interface over HTTPS, which replaced an older, less secure protocol and made fleet automation practical at scale. ## Why it is a security boundary A controller is small, independent, and able to power-cycle hardware, re-image a machine, and observe its console. That is complete physical control expressed as a network service, which makes the management network one of the most sensitive assets in a facility. The established practices follow directly: management interfaces on a physically or logically separate network reachable only through controlled paths, never from the internet; individual credentials rather than shared defaults; signed firmware with verified boot on the controller itself; and firmware updates treated with the same discipline as operating system patching. Treating the management plane as an afterthought gives an intruder the keys to the building; treating it as a first-class security domain is routine and effective. ## Relationships - **manages:** [[wiki/Host CPU|Host CPU]] · [[wiki/Compute Tray|Compute Tray]] · [[wiki/Accelerator|Accelerator]] - **reports to:** [[wiki/Observability|Observability]] · [[wiki/Site Reliability Engineering|Site Reliability Engineering]] - **enables:** [[wiki/Bare Metal Compute|Bare Metal Compute]] provisioning - **secured through:** [[wiki/Data Center Security|Data Center Security]] · [[wiki/Supply Chain Security|Supply Chain Security]] · [[wiki/Confidential Computing|Confidential Computing]] - **part of:** [[wiki/Rack and Cluster Integration|Rack and Cluster Integration]] - **source articles:** [[articles/Vertically Integrating an AI Superpower from AI Factory to Citizen|Vertically Integrating an AI Superpower from AI Factory to Citizen]] · [[articles/The Gigawatt Is the New Measure of War|The Gigawatt Is the New Measure of War]] - **collection route:** [[collections/American Infrastructure|American Infrastructure]] · [[collections/American|American Collection]] ## Evidence posture - **Established:** independent processor, firmware, power domain, and network path; power control, console redirection, virtual media, sensor telemetry, persistent event logging, firmware management, and hardware inventory; standardized management over HTTPS replacing an earlier protocol; the controller as a complete physical-control surface and the resulting management-network isolation, credential, and signed-firmware practices. - **Strongly indicated:** firmware attestation becoming standard in data center servers; capability ships broadly and operational adoption varies. ## Sources - DMTF Redfish specifications and server platform management practice, as current October 2026.