When evaluating a data center, one of the first questions is how reliable it actually is. The answer most often comes in the form of a Tier rating — a standardised classification developed by the Uptime Institute, a globally recognised authority on data center infrastructure.
Tier IV is the highest classification in the system. It represents a facility that can sustain any single equipment failure without disrupting operations — no interruptions during maintenance, no downtime from a component failing.
The Uptime Institute Tier classification
The Uptime Institute defines four tiers, each building on the one before it:
Tier I — Basic. A single path for power and cooling, no redundancy. Expected uptime: 99.671% (~28.8 hours of downtime per year). Used for small offices and non-critical workloads.
Tier II — Redundant Components. Redundant power and cooling components (N+1), still a single distribution path. Expected uptime: 99.741% (~22 hours per year). Maintenance still requires planned shutdowns.
Tier III — Concurrently Maintainable. Multiple power and cooling distribution paths, but only one active at a time. All equipment can be maintained without taking the site offline. Expected uptime: 99.982% (~1.6 hours per year of planned outages). Most enterprise-grade data centers are Tier III.
Tier IV — Fault Tolerant. Multiple active power and cooling paths simultaneously. The facility can sustain any single failure — a component failure, a power path failure, a cooling failure — without any impact on load. Expected uptime: 99.995% (~26 minutes of downtime per year, unplanned only; zero planned downtime).
What Tier IV means in practice
The defining characteristic of Tier IV is fault tolerance, not just redundancy. The difference is significant:
- Redundancy (Tier III) means a backup component exists. If the primary fails, operations may pause briefly while switching to backup, or maintenance on one path requires careful scheduling.
- Fault tolerance (Tier IV) means two fully active, independent paths are always running simultaneously. If one fails, the other carries the full load immediately — no manual intervention, no service interruption.
The engineering requirements
To achieve Tier IV, every component and system in the data center must meet these criteria:
2N redundancy minimum. For every critical component — power supply units, UPS systems, cooling units, generators — there must be at least double the required capacity, with fully independent paths. This means two separate utility feeds, two independent UPS systems, two sets of cooling equipment, and two separate distribution paths all the way to each rack.
Continuous cooling. Cooling cannot be interrupted for any reason. Tier IV facilities maintain continuous cooling even during the highest-demand maintenance operations.
No single point of failure. The entire infrastructure — from the utility connection through every distribution path to each individual rack — must be designed so that the failure of any single component has zero impact on customers.
Concurrent maintainability. Any component can be removed, replaced, or maintained at any time without affecting operations. Unlike Tier III, this is achieved through active redundancy, not just the existence of a backup.
Uptime numbers in context
The Tier IV target of 99.995% uptime translates to approximately 26 minutes of unplanned downtime per year — and zero planned downtime. For comparison:
| Tier | Uptime | Downtime per year |
|---|---|---|
| Tier I | 99.671% | ~28.8 hours |
| Tier II | 99.741% | ~22.0 hours |
| Tier III | 99.982% | ~1.6 hours |
| Tier IV | 99.995% | ~26 minutes |
The gap between Tier III and Tier IV is not a marginal improvement. A Tier III facility must schedule maintenance windows, and a failure during that window can cause an outage. A Tier IV facility eliminates this risk entirely.
Who actually needs Tier IV
Not every workload justifies the cost premium of Tier IV infrastructure. The cases where Tier IV is genuinely necessary:
Financial institutions and trading platforms. Even a minute of downtime can mean significant financial loss or regulatory penalties. Banking core systems, payment processors, and high-frequency trading infrastructure are classic Tier IV candidates.
Telecom and critical communications. Mobile network core, emergency services infrastructure, and national communications backbones cannot tolerate outages.
Government and defence. Agencies managing critical state systems — tax databases, registries, identity systems — require the highest available reliability guarantees.
Healthcare. Hospital information systems, clinical data repositories, and pharmaceutical manufacturing systems where downtime has direct patient safety implications.
Large-scale enterprise colocation. Organisations co-locating their core IT infrastructure expect a level of reliability that matches or exceeds what they could achieve in their own data center.
AI and HPC infrastructure. Long-running training jobs and inference workloads where an interruption means restarting hours or days of computation.
Tier IV vs Tier III: when Tier III is enough
Tier III is the right choice for most enterprise workloads. It provides 99.982% uptime — fewer than two hours of planned downtime per year — and concurrently maintainable infrastructure. For web applications, corporate IT, development environments, and secondary systems, Tier III delivers enterprise-grade reliability at a lower cost than Tier IV.
The decision to require Tier IV should be driven by a genuine business requirement: a regulatory mandate, a contractual SLA, or a documented business impact calculation showing that the cost of downtime exceeds the premium for Tier IV.
Certification and verification
An important distinction: self-claimed Tier rating vs. certified Tier rating.
The Uptime Institute issues formal Tier Certifications following a rigorous design review and operational inspection process. A data center can claim to be “designed to Tier IV standards” without formal certification — these claims are unverified.
When Tier IV reliability is business-critical, look for the official Uptime Institute Tier IV Certification — a document that confirms an independent third party has reviewed the design and operations and found them to meet the standard.
Tier IV in Kazakhstan
Akashi Data Center in Astana holds Tier IV Certification from the Uptime Institute — the first and only data center in Central Asia to hold this certification. The facility features:
- 2N+1 redundancy across all power and cooling systems
- Four independent fibre entry points from separate carriers
- Redundant 100 MW power infrastructure with independent UPS and generator sets
- Continuous cooling with no planned maintenance windows requiring shutdown
- Zero single points of failure across the entire infrastructure
For organisations that require the highest available reliability standard in Kazakhstan or Central Asia, Tier IV certification is the verifiable guarantee.
Frequently asked questions
What does 99.995% uptime mean for Tier IV?
It means the facility is engineered for no more than approximately 26 minutes of unplanned downtime per year. Additionally, Tier IV has no planned downtime at all — every maintenance operation is performed while the facility remains fully operational.
What is the difference between fault tolerance and redundancy?
Redundancy means a backup component exists and can be switched in if the primary fails. Fault tolerance means the backup is already active and takes load instantly if the primary fails — with no interruption. Tier IV is fault tolerant; Tier III is redundant but not fault tolerant.
Is Tier IV certification required or optional?
Formal Uptime Institute Tier IV Certification is optional — a data center can operate at Tier IV standards without going through the certification process. However, certification provides independent third-party verification. For regulated industries and demanding enterprise buyers, certified Tier IV is the standard of proof.
How much more expensive is Tier IV than Tier III?
Tier IV infrastructure costs 15–25% more to build than Tier III, due to the additional redundant systems. Colocation pricing in Tier IV facilities is typically higher than Tier III. The premium is justified for workloads where the cost of downtime exceeds the cost of the upgrade.
Can a data center upgrade from Tier III to Tier IV?
Yes, but it typically requires significant infrastructure investment: adding a second active distribution path for power and cooling throughout the facility. In some cases it is more cost-effective to build a new Tier IV facility than to retrofit an existing Tier III one.
Looking for Tier IV colocation in Central Asia? Learn about Akashi’s infrastructure — the region’s first and only Uptime Institute-certified Tier IV data center, located in Astana, Kazakhstan.