The Four Tiers of Availability

Data centers are categorized based on their redundancy and the amount of downtime expected per year.

Tier I: Basic Capacity

• Availability: 99.671% (~28.8 hours downtime/year).
• Design: Single non-redundant path for power and cooling. No backup components.
• Use Case: Small business server rooms where downtime has a minimal financial impact.

Tier II: Redundant Components

• Availability: 99.741% (~22.7 hours downtime/year).
• Design: Basic redundancy in components (N+1). Example: Two UPS systems where only one is needed for the load.
• Use Case: Regional offices or small local data centers.

Tier III: Concurrently Maintainable

• Availability: 99.982% (~1.6 hours downtime/year).
• Design: Multiple distribution paths for power and cooling, but only one path is active. Critially: You can perform maintenance on any component without shutting down the facility.
• Use Case: Enterprises and cloud providers.

Tier IV: Fault Tolerant

• Availability: 99.995% (~26.3 minutes downtime/year).
• Design: Multi-active paths everywhere (2N+1). If a catastrophe (fire/flood) hits one path, the other maintains the load automatically. No single failure can stop the facility.

Infrastructure Components

Achieving Tier III or IV requires more than just a battery. It involves:

1. Dual Utility Feeds: Power coming from two different substations.
2. ATS (Automatic Transfer Switches): Rapidly switching to generators when utility power fails.
3. CRAC/CRAH Units: Redundant cooling units to manage thermal load.
4. Water Storage: On-site water tanks to maintain cooling even if city water is cut.

Conclusion

Designing for high availability is an exercise in identifying and eliminating Single Points of Failure (SPOFs). While Tier IV is significantly more expensive, for businesses that lose thousands of dollars per minute of downtime, it is an essential investment.