What Is N+1 Redundancy?_
In data-center power distribution, N+1 redundancy means that for a given load requiring N units of capacity (e.g., power supplies, UPS modules, or cooling units), the system is equipped with N+1 units. This allows the system to sustain the full load even if one unit goes offline for maintenance or failure. It is a common design for critical infrastructure but does not cover simultaneous multiple failures.
Technical Details
N+1 redundancy is typically implemented at the rack level with redundant power supply units (PSUs) in servers or at the facility level with parallel UPS modules. The extra unit is often configured in a load-sharing or standby mode, and automatic failover must be tested to ensure no single point of failure. In GPU clusters, N+1 power redundancy is common for PSUs in the rack. The specific number of redundant units and their capacity ratings must follow the OEM spec for the power shelf or UPS.
How Leviathan Systems Works with N+1 Redundancy
During rack assembly, we verify that each GPU server and network switch has its PSUs configured for N+1 redundancy, and we label power cables accordingly. In liquid cooling commissioning, we check that the coolant distribution unit (CDU) has N+1 pump redundancy as per the design.
Related service
GPU Rack Assembly & Rack-and-Stack →