What Are Data Centre Tiers?
The Uptime Institute Tier Classification System is the global standard for evaluating data centre infrastructure reliability. The four-tier system rates facilities on redundancy and uptime. It is the single most referenced specification in colocation contracts and lease negotiations.
Higher tiers mean more redundancy at significantly higher cost. Tier III handles 90% of commercial use cases. Tier IV is overkill for anything that can tolerate a 2-minute failover.
Tier I: Basic Capacity
**Uptime SLA:** 99.671% (28.8 hours of downtime per year)
**Key characteristics:** - Single path for power and cooling distribution - No redundant capacity components - Must be shut down entirely for maintenance - 3-6 months to implement
**What this means in practice:** A Tier I facility has no backup systems. If any component fails or requires maintenance, the entire facility (or affected zone) goes offline. This is acceptable for development environments, non-critical testing, and data that can tolerate periodic outages.
**Construction cost:** ~$6-8M/MW **Typical lease rate:** $60-80/kW/month
Tier II: Redundant Capacity Components
**Uptime SLA:** 99.741% (22 hours of downtime per year)
**Key characteristics:** - Single path for power and cooling distribution - Redundant capacity components (N+1) - Engine generators and UPS modules can be removed for maintenance without shutdown - 6-12 months to implement
**What this means in practice:** Tier II adds backup UPS modules and generators, so individual components can be maintained without a full shutdown. However, the distribution path (switchgear, wiring) is still single-path. A failure in the distribution infrastructure still causes downtime.
**Construction cost:** ~$8-10M/MW **Typical lease rate:** $80-110/kW/month
Tier III: Concurrently Maintainable
**Uptime SLA:** 99.982% (1.6 hours of downtime per year)
**Key characteristics:** - Multiple active power and cooling distribution paths (only one active at a time) - Redundant capacity components (N+1) - Any component can be removed for planned maintenance without affecting IT load - 15-20 months to implement
**What this means in practice:** Tier III is the most common commercial data centre standard. The dual distribution paths mean the facility can be fully maintained without any planned downtime. The key word is "planned" - an unplanned failure in the active distribution path will still cause an interruption until the alternate path is activated.
This is the sweet spot for most enterprise workloads, colocation, and cloud infrastructure.
**Construction cost:** ~$11-14M/MW **Typical lease rate:** $110-160/kW/month
Tier IV: Fault Tolerant
**Uptime SLA:** 99.995% (0.4 hours of downtime per year)
**Key characteristics:** - Multiple active power and cooling distribution paths (simultaneously active) - Redundant capacity components (2N or 2N+1) - Facility can sustain any single infrastructure failure without affecting IT load - 15-20 months to implement
**What this means in practice:** Tier IV is true fault tolerance. Both distribution paths are simultaneously active, and the facility is designed so that any single failure (including a failure during maintenance of another component) does not affect the IT load. This is the highest level of reliability available.
Tier IV is required for mission-critical financial systems, real-time trading platforms, emergency services, and any application where even seconds of downtime have severe consequences.
**Construction cost:** ~$16-22M/MW **Typical lease rate:** $160-220/kW/month
Which Tier Do You Need?
| Workload | Recommended Tier | Why |
|---|---|---|
| Development/testing | Tier I-II | Non-production; outages are acceptable |
| General enterprise | Tier III | Balance of reliability and cost |
| E-commerce, SaaS | Tier III | Customer-facing but can tolerate brief planned maintenance windows |
| Financial trading | Tier IV | Milliseconds of downtime = millions of dollars |
| Healthcare/emergency | Tier IV | Regulatory requirements for continuous availability |
| AI training | Tier II-III | Long-running jobs can checkpoint and resume |
| Content delivery | Tier II-III | Geographic redundancy provides application-layer resilience |
Beyond Tiers: What Else Matters
Tier classification measures infrastructure redundancy, but it does not measure: - **Operational excellence**: A Tier IV facility with poor operating procedures can underperform a well-run Tier III - **Network redundancy**: Tier classification covers power and cooling only, not connectivity - **Security**: Physical and cyber security are separate from tier classification - **Sustainability**: PUE, renewable energy percentage, and water usage are independent of tier
Evaluate Facilities on Datacentres.com
Browse our facility directory to compare data centres by tier, capacity, and location. For help selecting the right tier for your workload, contact our advisory team.