Critical clinic inference servers need defined redundant PSU behavior under UCS. Where is that primarily configured?

Critical clinic inference servers need defined redundant PSU behavior configured in the UCS power policy that specifies how dual supplies should act for those AI servers. Redundancy mode and related power settings belong in policy so failover behavior is intentional, documented, and consistent across the clinic fleet rather than discovered during an outage. Domain 3.2.b covers that redundancy configuration for municipal AI hosts that must stay up through a supply fault. The UCS BIOS policy that defines redundant PSU power savings for those AI servers is not where redundant PSU behavior is defined; it only controls BIOS-level CPU, memory and power-performance settings and cannot enforce power-supply failover by UCS or Intersight. The UCS CIMC policy that defines redundant PSU restore state for those AI servers likewise creates no PSU behavior; management-controller and power-restore settings do not govern redundant power-supply failover. The UCS power capping policy that defines redundant PSU budget for those AI servers is not where redundant PSU behavior is defined; it governs power limits and budgeting and cannot configure power-supply redundancy. Exam caveat: physical dual-PSU cabling to separate PDUs is still required; the power policy defines logical redundant behavior once the hardware path exists. Operational check: set redundant PSU behavior in the UCS power policy for clinic inference servers, confirm Intersight/UCS shows the policy applied, validate status shows both supplies as expected in a maintenance window, and document the recovery expectation for a single PSU or PDU fault.