Data center tiers are a classification system, ascending 1, 2, 3, and 4 – with some operators even pushing for 5 – that are used to evaluate data center facilities, in a consistent way, regarding their potential site infrastructure availability, also known as uptime. Specifically, the tier ratings stipulate what a data center is able to offer in terms of redundancy and resiliency, as well as how much potential downtime a customer could experience over the course of a year.
As a general rule, the difference between data center tiers is that tier 1 offers no redundancy of any critical system, tier 2 has partial redundancy in their electrical & HVAC systems, tier 3 contains dual redundancy for power & cooling equipment, and tier 4 possesses fully redundant infrastructure.
Data centers are commonly rated by the Uptime Institute, an independent organization, which has issued over 2,500 certifications to data centers in more than 110 countries. The Uptime Institute ranks data centers through four distinct tier certification levels: Tier I, Tier II, Tier III, and Tier IV. In terms of availability, Tier I has the most expected downtime or worst performance, while Tier IV offers the least anticipated downtime or best performance.
Uptime Institute – Tier Standard
Importantly, the Uptime Institute’s four data center tiers are progressive, meaning each level includes the requirements of all the lower tiers.
What Factors Determine a Data Center’s Tier?
The Uptime Institute’s tier certification assigns ratings to data centers based on criteria for power, cooling, maintenance protocols, redundancy (i.e., duplicating critical components), and fault tolerance.
At the same time, these ratings are agnostic to the specific technologies and vendors that are supplying the facilities. Therefore, data centers are assessed purely on their capabilities and not on which manufacturer they source their electrical systems and HVAC components from (e.g., Schneider Electric, Eaton, and Vertiv).
How Do Data Centers Obtain a Tier Rating?
Data center tier certification ratings are not a regulatory requirement, meaning they are optional. Therefore, not all data centers have an assigned tier.
For a data center facility to obtain a tier certification from the Uptime Institute, the operator must send its site plans and engineering & architectural specifications, also known as Tier Certification of Design Documents, to the Uptime Institute for their review. Subsequently, as part of its assessment, the Uptime Institute’s personnel will visit the particular data center to inspect its operations, comparing how the facility aligns with its initial design documentation. Once its review is complete, the Uptime Institute will then assign a rating.
Data Center Tiers – What’s the Difference Between 1, 2, 3, 4, and 5?
Data center tiers, as classified by the Uptime Institute, include the levels of 1, 2, 3, and 4, while data center operator Switch Inc offers a proprietary tier 5 standard.
Note: for simplicity, Dgtl Infra will refer to data center tiers by their numerical values, as opposed to the Roman numeral syntax used by the Uptime Institute.
Summary Comparison of Data center Tiers 1, 2, 3, 4, and 5
|Category||Tier 1||Tier 2||Tier 3||Tier 4||Tier 5|
|Downtime (Annual)||<28.8 hours||<22.7 hours||<1.6 hours||<26.3 minutes||<5.3 minutes|
|Redundancy||None||Partial||N+1||2N or 2N+1||2N or 2N+1|
|Cost to Build||$||$$||$$$||$$$$||$$$$$|
While data center redundancy and resiliency are of critical importance, it also worth noting that there is a direct correlation between the tier level and the cost to build a data center.
Tier 1 Data Center – Basic Capacity
Tier 1 data centers provide dedicated space for IT infrastructure to support workloads beyond an office environment. This IT infrastructure includes an electrical backup generator, uninterruptible power supply (UPS), and HVAC / cooling systems, such as a computer room air conditioner (CRAC) unit. Notably, these facilities have a single distribution path for power and cooling infrastructure, and there is no redundancy of any critical system.
Collectively, these basic data center components enable the facility to manage through power spikes and outages, as well as operate outside of typical office hours – meaning nights and weekends. However, in order to perform regular maintenance or emergency repairs, the data center operations must be shut down temporarily.
Tier 1 data centers are the lowest-rated tier, only guaranteeing 99.671% uptime per year, implying a maximum total annual downtime of 1,729 minutes, which is equivalent to 28.8 hours.
Given a tier 1 data center’s limitations, they are best suited for very small enterprises seeking the most cost-effective hosting solution. Additionally, these enterprises need to be able to tolerate meaningful periods of downtime over the course of a year (i.e., up to 28.8 hours annually).
Tier 2 Data Center – Redundant Capacity
Tier 2 data centers contain all of the capabilities of tier 1 facilities, but also include partial redundancy in the form of electrical backup generators, power components (e.g., UPS systems), and cooling equipment (e.g., chillers / pumps). In turn, redundant infrastructure enables tier 2 data centers to undertake select maintenance opportunities, while providing an increased level of protection against IT process disruptions resulting from equipment failures.
Similar to tier 1 facilities, tier 2 data centers rely on a single distribution path for power and cooling, which makes these facilities vulnerable to unexpected disruptions.
Tier 2 data centers guarantee 99.741% uptime per year, implying a maximum total annual downtime of 1,361 minutes, which is equivalent to 22.7 hours.
Tier 2 data centers are commonly used by small- to medium-sized enterprises that want a cost-effective and more reliable option for their compute and storage needs. Typically, these enterprises use tier 2 facilities for data backup & recovery purposes, as well as non-mission critical functions.
Additionally, tier 2 data centers satisfy the requirements of customers who have connected together their data centers or can afford a reasonable amount of downtime (i.e., up to 22.7 hours annually). Frequently, tier 2-rated data centers are designed close to tier 3 standards, however, the operator may have chosen to reduce costs by fitting-out the facility at tier 2 levels for non-critical components.
Tier 3 Data Center – Concurrently Maintainable
Tier 3 data centers contain all of the capabilities of tier 1 and 2 facilities but do not require shutdowns during planned maintenance and equipment replacement, hence they are concurrently maintainable. To implement this resiliency, an additional redundant distribution path for power and cooling is added to the existing redundant critical components of tier 2 facilities so that all components needed to support the IT processing environment can be shut down and maintained, without impacting IT operations.
More specifically, all IT equipment must have dual power supplies attached to different UPS units, ensuring that a UPS unit can be taken offline, without causing servers to crash or creating a loss in network connectivity. Also, redundant cooling systems must be in-place such that if one cooling unit fails, the other one starts-up and continues to moderate the temperature and humidity in a data center’s computer rooms.
By being concurrently maintainable, tier 3 data centers have systems, protocols, and equipment in place to permit staff to perform maintenance and repairs without interrupting services to customers.
Tier 3 data centers guarantee 99.982% uptime per year, implying a maximum total annual downtime of 94.6 minutes, which is equivalent to 1.6 hours.
Tier 3 data centers offer more comprehensive redundancy capabilities and a significant improvement in availability. These facilities offer N+1 redundancy, meaning that an additional component will start operating for the purposes of supporting a single failure or planned maintenance on a component. Below are the definitions of N+1 redundancy:
- N: necessary capacity in order to support the full IT load of the data center
- +1: extra component for backup purposes
Also, tier 3 data centers require backup solutions which can keep operations running, for at least 72 hours, following a local or region-wide power outage.
Not Fault Tolerant
Tier 3 data centers are not entirely fault tolerant because they often rely on, or share components, which are not completely independent to the facility. For example, these components could include utility company power feeds and external cooling systems that reside outside the data center.
Tier 3 data centers are the industry standard, providing customers with high-quality space, power, and cooling, which delivers a sufficient level of redundancy and resiliency to support a customer’s mission-critical compute and storage needs. As such, tier 3 data centers are preferred by larger enterprises, particularly if they host and manage customer data.
Tier 4 Data Center – Fault Tolerant
Tier 4 data centers contain all of the capabilities of tier 1, 2, and 3 facilities, but also include fault tolerance mechanisms, with redundancy for every component. Specifically, fault tolerance means that when any unplanned individual equipment failures or distribution path interruptions occur, the effects of the events do not impact IT operations.
Essentially, tier 4 facilities have no single points of failure. Therefore, the objective of a tier 4 data center is to provide continuous support and services no matter the circumstances.
Tier 4 data centers are the highest-rated tier (by the Uptime Institute), guaranteeing 99.995% uptime per year, implying a maximum total annual downtime of 26.3 minutes, which is equivalent to 0.4 hours.
2N or 2N+1 Redundancy
Tier 4 data centers either have 2N or 2N+1 redundancy, meaning fully redundant infrastructure – which is the main difference between tier 3 and 4 facilities. Particularly, 2N redundancy means there is a completely identical system on standby for every component, with this system being physically isolated and independent from the primary system. In turn, this physical separation prevents a local event from compromising both systems.
More specifically, all components are supported by two different utility power suppliers, two electrical backup generators, two UPS systems, two power distribution units (PDUs), and two different cooling systems.
Furthermore, each distribution path is independent, such that a single failure in one does not cause a chain reaction of failures with other components. If any individual power or cooling infrastructure component fails in a tier 4 data center, processing will continue without issue. Therefore, IT processing can only be impacted if components from two different electrical or cooling paths fail.
Also, tier 4 data centers require backup solutions which can keep operations running, for at least 96 hours, following a local or region-wide power outage. Finally, the facility’s power source does not connect to any external source, meaning it must be independent.
Cost to Build
Given the fully redundant requirements of a tier 4 data center, the costs of fitting-out such a facility are much higher and, as a result, tier 4 data centers remain relatively rare. For example, the cost of constructing and fitting-out a tier 4 data center can be 25% to 40% more than a tier 3 data center and double that of a tier 2 data center.
Tier 4 data centers serve the needs of enterprises who are not price-sensitive and require uninterrupted availability. For example, large enterprises (e.g., financial services) with mission-critical workloads and demanding customer or business needs are often clients of tier 4 data centers.
Tier 5 Data Center – Fault Sustainable
Tier 5 data centers go beyond the Uptime Institute’s tier certification standards. Specifically, Switch Inc, a retail colocation provider, introduced a proprietary Tier 5 Platinum standard, which attempts to set a level above fully redundant systems and total fault tolerance. This standard incorporates more than 30 additional elements critical to data center design and constant operation.
Switch’s Tier 5 Platinum standard includes rigorous parameters regarding long-term power system capabilities, the number of available on-net carriers, the durability of roof systems, the location of cooling system lines in or above the data center, physical & network security, and 100% use of renewable energy.
Why Do Data Centers Get Tier Certifications?
The Uptime Institute’s tier certifications provide a quick reference point to determine a data center’s level of performance. For data center operators, these ratings often deliver benefits in the following ways:
- Marketing: operators can show existing and prospective customers that their data center meets credible standards and can support the needs of enterprises, as determined by an independent third-party
- Customers: data centers rated tier 3 or higher may meet the criteria of a wider audience of prospective customers, who internally set minimum certification requirements. This is particularly important for newer data center operators, which do not have a pre-existing relationship with a particular enterprise
- Planning: certification assessments deliver guidance on addressing deficiencies, with guidelines for making future improvements, upgrades, and facility expansions
Ultimately, data center tiering classifications are used as one component in determining the quality of a data center. Inevitably, customers of data centers have internal operating and availability requirements that they expect of any facility. In particular, larger enterprises, such as multinational corporations, as well as carriers have their own standards and requirements for evaluating data centers.
Given that data center tier certification ratings are not a regulatory requirement, the design of a data center is generally determined by the commercial agreement between a data center operator and its customers – which tend to be unique to the needs of a specific customer. As such, many data centers do not pursue a formal tier certification process.