NVIDIA Rack Management Service

View as Markdown

The Rack Management Service (RMS) is a stateless, rack-level hardware management service for NVIDIA data-center infrastructure. It exposes a single gRPC API that trusted services and operators use to run power control, inventory, firmware updates, and switch configuration against targets ranging from a full rack down to an individual component.

RMS translates client intent into hardware actions across multiple management protocols: it talks to compute BMCs and power-shelf controllers over Redfish/HTTPS, and to NVSwitch trays over NVUE REST, SSH, SFTP, and NMX-C gRPC. Long-running work (firmware updates, switch system-image installs) runs as asynchronous jobs that clients poll to completion.

Where to Go Next

Run & Operate RMSIntegrate with RMSEvaluate RMS
WhoOperators deploying and running RMS against real racksPlatform engineers building on the RMS gRPC APIArchitects evaluating RMS for their stack
Start hereGetting Started: PrerequisitesArchitectureOverview
ThenDeploymentOperations (RPC reference)Hardware Compatibility List
ThenConfigurationDevelopmentGlossary