Data centre tiers are a four-level system, ranging from Tier I to Tier IV, that explains how reliably a data centre can keep running during maintenance or unexpected equipment failures. Developed by the Uptime Institute, the system helps businesses understand how well a facility protects their IT equipment from downtime. Tier III means any component can be maintained without shutting down your servers, and Tier IV adds tolerance of an unplanned fault. If you are shortlisting colocation in the UK, the tier tells you a lot about resilience, but only if you know what was actually assessed. This guide explains each tier, the difference between Tier 3 and Tier 4, where the familiar uptime percentages come from, and how to check a provider's claim.
Key takeaways
- The Uptime Institute defines four tiers: Tier I basic capacity, Tier II redundant capacity components, Tier III concurrently maintainable, Tier IV fault-tolerant.
- The main difference between Tier 3 and Tier 4 is that Tier III survives planned maintenance without downtime, while Tier IV is also designed to stop an unplanned failure from reaching your IT.
- The 99.671%, 99.741%, 99.982% and 99.995% uptime figures are not part of Uptime Institute's current standard, which dropped expected downtime per year in 2009.
- Only the Uptime Institute can certify a tier, and it certifies designs and built facilities separately, so ask which certificate sits behind any claim.
What are data centre tiers?
The Uptime Institute's Tier Classification System measures how well a data centre's essential infrastructure, including power and cooling systems, can cope with maintenance and equipment failures. It helps businesses understand a facility's resilience and how likely it is to experience downtime. Each tier includes everything in the tier below it.
Each tier describes how reliably a data centre operates, rather than simply how much equipment it has. A higher tier isn't automatically the right choice for every business. Uptime Institute states in its explanation of the Tier Classification System that the tiers are progressive but that "a Tier IV solution is not always 'better' than a Tier II solution". The right tier depends on what the workload is worth to your business.
Tiers say nothing about security, connectivity or energy efficiency. A Tier III site can have a poor power usage effectiveness score, which is why PUE is a separate question to ask any provider.
What is a Tier 1, 2, 3 and 4 data centre?
The four data centre tiers build on each other, from basic capacity with no redundancy to a fully fault-tolerant design.
| Tier | Uptime Institute name | What it means in practice | Planned maintenance | Unplanned failure |
|---|---|---|---|---|
| Tier I | Basic capacity | Dedicated IT space, UPS, dedicated cooling and an engine generator | Needs a full shutdown | Likely to cause an outage |
| Tier II | Redundant capacity components | Adds spare power and cooling units, such as UPS modules, chillers or pumps | Some units can be serviced without a shutdown | A distribution path failure still causes an outage |
| Tier III | Concurrently maintainable | Adds a redundant delivery path for power and cooling | Any component can be maintained with no shutdown | Can still cause an outage, especially during maintenance |
| Tier IV | Fault tolerant | Builds on Tier III so individual failures are stopped short of the IT | No shutdown | Designed to keep IT running |
Tier II protects against a failed unit but not a failed cable route, switchboard or pipe, so most colocation buyers end up comparing Tier III and Tier IV.
What is a Tier 3 data centre?
A Tier III data centre, often written as Tier 3, is concurrently maintainable. Every component that supports your IT can be shut down for maintenance or replacement without shutting down the IT itself. Uptime Institute's own wording is that a Tier III facility "requires no shutdowns for equipment replacement and maintenance".
This is possible because the data centre has backup equipment and alternative routes for delivering power and cooling. This means engineers can maintain one part of the system while another keeps the IT equipment running.
Spare capacity alone is not enough. A site with N+1 UPS modules feeding a single distribution board still has a single point of failure at that board. We cover how unit-level redundancy works in our guide to what N+1 redundancy means, and why it is a different measure from a tier.
What is meant by Tier 4 data centre?
A Tier IV data centre is fault-tolerant: when an individual piece of equipment fails, or a distribution path is interrupted, the effect is stopped before it reaches the IT. Tier IV builds on everything in Tier III and adds that protection against unplanned events.
Tier IV data centres use separate, independent power and cooling systems to reduce the risk of one problem affecting everything. For example, if one power system fails, the other can continue supplying the equipment. The Uptime Institute's description of Tier IV also requires continuous cooling, so the IT space stays within limits during the switchover from a failure.
Tier IV is defined by outcome rather than by unit counts. The tier does not simply mean "2N everything", and adding more equipment does not, on its own, move a site up a tier.
Tier 3 vs Tier 4 data centre: what is the difference?

The main difference between Tier 3 and Tier 4 is how they handle unexpected problems. Tier 3 allows maintenance to take place without shutting down the IT equipment. Tier 4 goes a step further, keeping systems running even if an individual piece of equipment or power route fails.
| Comparison | Tier III | Tier IV |
|---|---|---|
| Core idea | Concurrently maintainable | Fault tolerant |
| Planned maintenance | No IT shutdown | No IT shutdown |
| Single unplanned equipment or path failure | May affect IT | Stopped before it reaches IT |
| Physically separated, independent systems | Not a defining requirement | Required |
| Typical buyer | Most business and production workloads | Workloads where any interruption carries serious cost or risk |
| Build and running cost | Lower | Higher, because more plant and space sit idle as protection |
Tier IV costs more to build and operate because more infrastructure exists only to absorb failures, and that cost flows through to colocation pricing. Decide whether your workload needs fault tolerance or whether concurrent maintainability plus your own application-level resilience is enough.
How much downtime does each tier allow?
Uptime Institute's current Tier Standard does not assign an availability percentage or allowed downtime to any tier. The Uptime Institute states that in 2009 it "removed all references to 'expected downtime per year'", and that the current Tier Standard "does not assign availability predictions to Tier levels".
The percentages you see on most tier guides come from Uptime Institute material published before that change, and are still widely quoted. To put these percentages into perspective, here's how much downtime each figure would represent over a typical 365-day year:
| Tier | Legacy availability figure | Equivalent downtime per year |
|---|---|---|
| Tier I | 99.671% | About 28.8 hours |
| Tier II | 99.741% | About 22.7 hours |
| Tier III | 99.982% | About 1.6 hours (95 minutes) |
| Tier IV | 99.995% | About 26 minutes |
Treat these as rough historical benchmarks, not promises. Real availability depends on how a site is built, operated and maintained, so the figure that matters is the availability commitment in the contract and service level agreement (SLA), not a percentage attached to a tier label.
Does a data centre need Uptime certification to claim a tier?
A data centre needs Uptime Institute certification to say it is Tier certified, because the Uptime Institute is the only organisation that can formally certify against its Tier Classification System. Without a certificate, a provider can say its site is "built to Tier III" or "Tier III aligned", but that is the provider's own assessment.
Certification vs design
Uptime Institute certifies in stages, and each stage proves something different:
- Design certification: Confirms that the data centre's plans meet the requirements for its chosen tier.
- Constructed facility certification: Checks that the finished building matches the approved plans through on-site inspections and testing.
- Operational sustainability certification: Assesses how effectively the data centre is managed, maintained, and operated.
The gap between design and build matters. The Uptime Institute reports that, in its experience, "more than 85% of all data centre designs are incorrectly executed in construction". A design certificate alone does not prove the hall your servers sit in performs as drawn.
Other frameworks exist too. The European EN 50600 series uses availability classes 1 to 4, and the US standard ANSI/TIA-942 uses Rated 1 to Rated 4. These are separate standards, so a "Rated 3" or "Class 3" site has not been certified by the Uptime Institute.
Is Tier 3 good enough for my business?
Tier III is good enough for most business and production workloads, because it removes planned maintenance as a cause of downtime.
Tier IV is worth paying for when a single unplanned interruption would cause serious financial, safety or regulatory harm and your application cannot fail over to another site. For many organisations, a Tier III facility combined with replication to a second location gives better value.
A worked example
Say you are moving 10 production racks from an office server room in the South West, and your finance director asks whether you need Tier IV. Start with three questions. What does an hour of downtime cost you? Can the application fail over to a second site or cloud region? Do your regulators or customers name a specific tier? If downtime is costly but you can fail over, a concurrently maintainable facility plus a second site usually beats paying for fault tolerance in one place.
Questions to ask a colocation provider about its tier

Ask these questions before you sign:
- Is the site Uptime Institute Tier certified, and can I see the certificate?
- Is the certificate for design documents, the constructed facility, or operational sustainability?
- Does the tier apply to the whole facility or only certain halls?
- If the site is not certified, which tier is it designed to, and who assessed that?
- Do my racks get A and B power feeds from separate distribution paths?
- What availability does the service level agreement commit to, and what happens if it is missed?
Our air-cooled environments are built to N+1 redundancy across power and cooling, and dual power feeds and UPS systems are standard across hosted rack positions. You can review the standards we follow and ask us for supporting documentation.
Frequently asked questions
What is concurrent maintainability?
Concurrent maintainability means any component supporting the IT, including UPS units, chillers, switchboards and pipework, can be taken out of service for planned maintenance or replacement without shutting down the IT load. It is the defining feature of a Tier III data centre and relies on both spare capacity and a redundant distribution path.
Is Tier 3 the same as N+1?
No. N+1 means a data centre has at least one extra piece of equipment available if another unit needs maintenance or stops working. Tier III is broader, measuring whether the entire power and cooling infrastructure can be maintained without shutting down the IT equipment. A site can have N+1 UPS modules and still need a shutdown to service a single shared switchboard, so N+1 alone does not make a facility Tier III.
Do tiers apply to high-density AI racks?
Tiers apply to the site infrastructure serving any rack, including liquid-cooled AI racks. For high-density systems, check that the cooling distribution, such as coolant distribution units and pipework, is covered by the same redundancy as the power. Ask whether the tier claim covers the liquid-cooled halls or only air-cooled space.
Choosing the right data centre tier for your colocation
Data centre tiers tell you how a facility behaves during maintenance and failure: Tier III keeps your IT running through planned work, and Tier IV through a single unplanned fault. Ask for the certificate and the SLA rather than relying on a label.
If you need a private, controlled space for regulated or sensitive workloads, our secure cages and private halls connect to the facility's N+1 power infrastructure, with dedicated circuits for higher draw. To check whether your racks fit, book a power assessment, and our engineers will talk through power, cooling and resilience for your deployment.





