What is N+1, 2N, 2N+1 Redundancy in a Data Center

When you buy a UPS for your office, one is enough. When you buy infrastructure for a data center, the question changes: what happens if this component fails in production?

The N+1, 2N, and 2N+1 redundancy architectures answer that question. They are how the industry quantifies how much backup capacity a critical system has.

The base unit: N (the minimum capacity to operate)

N is the minimum number of components for the system to perform its function under normal load. If your UPS must deliver 100 kW to sustain the load, an N system is a 100 kW UPS. If the UPS fails, the load drops. Period.

That is the design of a basement with one UPS and one AC unit. It works for an office. For a DC operating 24×7, it is an outage waiting to happen.

N+1: one extra component over the minimum

N+1 adds one redundant component to the N system. If N is two 100 kW UPSs (total load 100 kW, one in passive standby), N+1 is two 100 kW UPSs plus a third 100 kW UPS. Any UPS can fail and the other two keep the operation running.

The core idea of N+1: tolerance to one failure. Not two at once. It is the most common architecture in small and medium data centers, and it shows up in UPSs, chillers, chilled water pumps, generators, and network switches.

2N: two complete, independent systems

2N means having two complete N systems, each capable of sustaining the load on its own, operating in parallel or in active-passive. If path A fails, path B keeps the operation running without visible interruption.

The classic example: two 100 kW UPSs each operating at 50%, fed from two different electrical substations, with two physically separated cabling routes. Any failure in one path does not affect the other.

2N is the foundation of Uptime Institute Tier III and Tier IV. It is what enables a service provider to sign availability SLAs of 99.982%.

2N+1: two complete systems plus one extra component

2N+1 adds one extra unit to the 2N system. If the two systems are two 100 kW UPSs each (load 100 kW), 2N+1 is three 100 kW UPSs (one in passive standby on top of the 2N architecture).

In practice, 2N+1 lets you perform scheduled maintenance on one of the 2N systems without affecting operational redundancy. The full system remains 2N while the extra component covers the maintenance.

When each level applies

The choice is not free. Each level increases CAPEX and OPEX. Moving from N+1 to 2N doubles the amount of equipment, maintenance contracts, and room space. That is why it is chosen based on the criticality of the service.

• N: offices, telecom racks without a formal SLA, non-critical loads.

• N+1: small and medium data centers, backup sites (DRP), internal company operations.

• 2N: service provider data centers with commercial SLA, financial institutions, regulated mission-critical sites.

• 2N+1: Tier IV data centers and environments where hot maintenance is performed on the infrastructure.

Relationship with the Uptime Institute Tier classification

Uptime Institute defines four Tier classification levels. Redundancy sits at the heart of each level, although Tier is not exactly the same as N+1 or 2N.

• Tier I: basic capacity, no redundancy (it is an N system).

• Tier II: N+1 redundant capacity in key components (UPS, chillers).

• Tier III (Concurrent Maintenance): Requires 1 active path and 1 backup path with N+1 components. It does not strictly require active-active 2N, but it allows maintenance on any component without shutting down the load.

• Tier IV (Fault Tolerance): Requires a fully independent, redundant, active-active 2N or 2N+1 architecture across all its paths and components.

Common mistakes when specifying redundancy

The most expensive mistake is specifying 2N and deploying N+1 because of budget or space constraints. This leaves the DC with an SLA signed for an architecture that does not physically exist, and an outage waiting for the first real failure.

• Confusing component redundancy with system redundancy. Two UPSs in parallel are not 2N if they share the same electrical substation and the same distribution panel.

• Specifying N+1 when the critical load requires 2N. An outage costs more than the initial savings on the second system.

• Forgetting that redundancy must also live in the cabling. Redundant UPSs with cabling in the same conduit are not 2N.

• Provisioning N+1 capacity on modular UPSs without verifying that a module failure does not degrade the entire system.

💡Redundancy is not a number; it is an architecture. Before signing an SLA or designing a DC, it pays to verify that each critical subsystem — power, cooling, network, storage — meets the declared redundancy level. Everything else is marketing.

Sources

[1] Uptime Institute — Tier Classification System (official definition) — https://uptimeinstitute.com/resources/asset/tier-classification-system

[2] TIA-942-C — Telecommunications Infrastructure Standard for Data Centers — https://tiaonline.org/resource/tia-942-c-data-center-infrastructure-standard/