Diagram of N+1, N+2, and 2N redundant chiller plants for data center cooling

Modular, Edge and Redundant Data Center Cooling: N, N+1, N+2 and 2N Explained

November 9, 2026

By Paul Johnson, Technology & R&D, G&D Chillers

Short answer: Redundant data center cooling means installing more chiller capacity than the load needs, so a unit can fail or be serviced without raising server temperatures. N+1 adds one spare chiller and is the most common design; 2N duplicates the entire plant for the highest uptime, and modular chillers make either approach easier to build and expand.

This guide compares redundancy levels with worked examples, explains when modular and edge cooling make sense, and shows how G&D's data center cooling platforms support each approach.

What do N, N+1, N+2 and 2N mean in data center cooling?

"N" is the number of chillers needed to carry the full design cooling load. Every redundancy level adds capacity on top of N.

Configuration What is installed Survives Typical use
N Exactly the capacity needed Nothing, any failure reduces cooling Non-critical rooms only
N+1 N units plus one spare One chiller failure or service outage Most enterprise and colocation sites
N+2 N units plus two spares One failure during planned maintenance Large colocation, AI campuses
2N Two complete, independent plants Loss of an entire plant or distribution path Mission-critical, financial, government
2(N+1) Two independent N+1 plants Plant loss plus a unit failure Highest-availability facilities

The Uptime Institute's Tier Standard ties these to site ratings: Tier III requires concurrently maintainable systems, which typically means at least N+1, while Tier IV requires fault-tolerant systems, commonly delivered with 2N distribution.

Diagram comparing N, N+1, N+2 and 2N chiller redundancy configurations

How much chiller capacity does each redundancy level require?

Redundancy is cheapest when the load is split across more, smaller chillers. The same 1,500 kW load needs very different installed capacity depending on unit size.

Worked example: 1,500 kW design cooling load

Configuration Unit size Units installed Installed capacity Overbuild
N+1 1,500 kW 2 3,000 kW 100%
N+1 750 kW 3 2,250 kW 50%
N+1 500 kW 4 2,000 kW 33%
N+2 500 kW 5 2,500 kW 67%
2N 500 kW 6 (two plants of 3) 3,000 kW 100%

With N+1, the more modules you use, the less spare capacity you pay for. Get your N first with our cooling requirements guide.

"Redundancy isn't only about the spare chiller. Check the pumps, the power feeds, the controls and the piping. We see plenty of N+1 chiller plants sitting behind a single header valve that makes the whole thing N." — Paul Johnson, Technology & R&D, G&D Chillers

Is N+1 or 2N better for a data center?

N+1 is the right choice for most enterprise and colocation data centers because it protects against a single chiller failure at far lower cost. 2N is justified when a cooling outage costs more than a second plant, or when tenants or regulators require fault tolerance.

Use these questions to decide:

  1. What does one hour of downtime cost the business or its tenants?
  2. Does a customer SLA or certification require Tier III or Tier IV?
  3. Can maintenance be scheduled only in low-load periods, or must it happen any time?
  4. Are there single points of failure elsewhere, such as power, pumps or piping, that would make chiller-only 2N wasted money?

What is modular data center cooling?

Modular data center cooling builds the plant from repeatable, factory-built chiller units that are added as IT load grows. Instead of installing the final plant on day one, operators add capacity in blocks that match tenant or GPU deployments.

Benefits of a modular approach:

  • Lower upfront cost. Buy capacity when revenue arrives, not years ahead.
  • Built-in redundancy. Each added module can become the "+1".
  • Faster deployment. Factory-tested units shorten commissioning.
  • Right-sized efficiency. Chillers run closer to their best part-load points.

G&D's GDX Air fits in standard 20 ft or 40 ft high cube containers and scales from 250 kW to 2,000 kW per unit, so modules can be staged across a campus. G&D's Modular Series applies the same expansion-module approach at smaller capacities.

What cooling works best for edge and modular data centers?

Edge data centers are small, distributed and often unstaffed, so the best cooling is simple, compact and remotely monitored. Capacities are usually well under 1 MW, which favors scroll or small oil-free chillers over large central plants.

Site type Typical IT load Recommended platform Why
Edge, telecom, network operations centers 20 to 300 tons G&D Scroll Chillers (tandem scroll, staged capacity) Simple service, staged redundancy inside one unit
Modular or containerized data center 250 kW to 2 MW per block GDX Air Container footprint, oil-free, high ambient rating
Phased colocation campus Multi-MW, added in blocks GDX Air or Air-Cooled Screw Repeatable modules for phased growth
Cool-climate modular build 250 kW to 2 MW per block GDX Free Integrated free cooling lowers PUE

For unstaffed edge sites, add remote monitoring so alarms reach an operator before temperatures climb. G&D offers remote monitoring for its chillers.

What else needs redundancy besides the chillers?

Chiller redundancy only works if the rest of the cooling path matches it. Review each of these against your redundancy target:

  • Chilled water pumps (N+1 at minimum)
  • Electrical feeds to chillers and pumps, including generator and UPS coverage for controls
  • Piping: loops or dual headers with isolation valves so any unit can be removed
  • Controls: no single controller that can stop the whole plant
  • How the plant recovers after a power interruption, plus thermal storage if needed to bridge it
  • CDUs and fan walls for liquid-cooled or high-density halls

Frequently asked questions

What is N+1 redundancy in data center cooling? N+1 means installing one more cooling unit than the load requires, so any single unit can fail or be serviced without losing cooling capacity.

What is the difference between N+1 and 2N? N+1 adds one spare unit to a single plant. 2N builds two complete, independent plants, each able to carry the full load alone.

What Uptime Tier needs N+1 cooling? Tier III requires concurrently maintainable infrastructure, which is typically met with at least N+1 cooling. Tier IV requires fault tolerance, commonly met with 2N.

Are modular chillers good for data centers? Yes. Modular chillers let operators add capacity as load grows, reduce the cost of redundancy and shorten deployment with factory-tested units.

What is the best cooling for an edge data center? Compact, air-cooled chillers with staged compressors and remote monitoring, such as scroll chillers from 20 to 300 tons, suit most edge sites.


Design a redundant, scalable plant. Explore G&D's data center platforms or read our introduction to N+1 redundancy.