Availability classes in the data centre: the basis of IT infrastructure __

Availability classes in the data centre: the basis of IT infrastructure
8. August 2025 8 min.

Today, dependence on functioning digital infrastructure is one of the most significant economic and risk factors – and the requirements for data centre availability classes are correspondingly high. Companies, public authorities, medical facilities, critical infrastructure operators – they all face the same pressure: any unplanned failure of digital systems can bring operational processes to a standstill or even endanger human lives.

The figures speak for themselves. According to studies, just one hour of downtime costs 98% of companies more than £100,000. In large corporations, the damage caused by downtime quickly runs into millions – per hour. The average damage worldwide is around £9,000 per minute. In highly regulated or systemically important sectors, this figure can be significantly higher.

The increasing complexity of digital processes, the trend towards real-time processing and competition for immediate customer access are further exacerbating this dependency. Uptime is therefore becoming a basic business requirement – not only in Frankfurt, but also in Singapore, Dallas and São Paulo.

This makes the question of data centre availability a strategic one. Anyone planning or operating infrastructure today must be able to demonstrate how fail-safe and redundant the system is. So-called "data centre availability" is becoming a key performance indicator – comparable to a credit rating in finance or a CO₂ footprint in ESG assessment.

Orientation in a billion-dollar market

The global market for data centres is expected to grow to over one trillion US dollars by 2034. In this environment, availability classes help to strategically align investments: they structure requirements, make operational risks measurable and create a common language between IT, architecture, operations and management.

But what exactly is behind availability classes AC1 to AC4 according to EN 50600 and Tier I to Tier IV according to the Uptime Institute? Which specific redundancy models – N, N+1, 2N or 2N+1 – are behind them? And why have these systems become international benchmarks for reliability and digital resilience?

 

Classification and significance: What do availability classes tell us?

Standardised availability is not a technical detail, but a planning tool. Availability classes are standard-based categories that can be used to classify the operational readiness of data centres – regardless of specific technologies or providers. They do not indicate how a data centre is specifically structured, but rather the operational reliability and fault tolerance that its design should achieve.

The basis is the European standard EN 50600 (also: DIN EN 50600) or the internationally recognised Tier model of the Uptime Institute. Both models follow a four-level system – VK1 to VK4 (or AC 1–4) and Tier I to IV – and define target values for redundancy, maintainability and fault tolerance.

Comparability instead of speculation

The real value of availability classes lies in their standardisation. While terms such as "highly available" or "redundant" were used inconsistently in the past, EN 50600 and the Tier models now provide an objective framework. This creates transparency:

  • For operators who want to systematically develop their infrastructure
  • For planners who need to tailor technical concepts to target levels
  • For customers who want to protect themselves – whether against data loss, reputational damage or regulatory risks

A key advantage: each level – whether Tier or VK – represents a clearly defined level of availability that can also be translated into downtime per year. The models therefore enable internationally consistent evaluation, regardless of whether a data centre is located in Europe, Asia or North America.

Why redundancy does not automatically mean availability

Redundant components alone do not guarantee high data centre availability. Only when the entire infrastructure is consistently designed for reliability – including power supply, cooling, network connection and security concept – can a resilient level be achieved.

Availability classes help to structure precisely this complexity. They replace technical gut feeling with a comprehensible target architecture. EN 50600 forms the European framework and is also the basis for the international standard ISO/IEC 22237.

Differences between EN 50600 and the tier model

Both systems pursue a similar goal, but differ in their assessment approach and scope:

EN 50600 (VK1–VK4):
– Focus on design architecture
– Assessment via planning, documentation and infrastructure concept
– Holistic approach: also includes energy efficiency, security zones, operational management
– Considered the European reference and basis for ISO/IEC 22237

Tier system (Tier I–IV):
– Focus on specifically verifiable fault tolerance
– Measurable uptime values for each level: e.g. Tier III with 99.982% (1.6 hours downtime/year), Tier IV with 99.995% (26 minutes)
– Over 3,500 certificates issued worldwide
– De facto standard for international operators – especially in North America and Asia

EN 50600 dominates in Europe, partly because it can be certified by independent bodies such as TÜV. Internationally, the Uptime Institute is the market leader for Tier certificates. Many operators now pursue a dual certification strategy: VK certification for regional regulations, Tier certification for international customer requirements.

The four levels of availability according to EN 50600 and Tier

The two leading systems for classifying data centres – EN 50600 and the Uptime Institute's Tier model – differ in their methodological approach but lead to comparable levels of reliability in terms of content. One system is design-based, the other is key figure-driven. Both enable a structured classification of availability and form the basis for strategic infrastructure decisions.

The following tables show a direct comparison of how Tier levels and availability classes (AC) differ in terms of availability, redundancy model and typical areas of application.

Availability levels according to the tier model (Uptime Institute)

 Tier level   Availability (year)  Downtime (year) Redundancy model   Typical application 

 Tier I 

99,671 %

 Approx. 28.8 hours 

 Single supply, no redundancy 

Small businesses, development, test environments

 Tier II 

 99,741 % 

Approx. 22 hours 

Partial redundancy (N+1), one supply path 

 SMEs, internal services, non-critical systems 

 Tier III 

 99,982 % 

Approx. 1.6 hours

 N+1, multiple paths, maintainable during operation 

 Enterprise IT, SaaS, e-commerce, healthcare 

 Tier IV 

 99,995 % 

 Approx. 26 minutes 

 2N or 2(N+1), fully fault-tolerant 

 Banks, stock exchanges, critical national infrastructure 

 

Availability classes according to EN 50600 (VK1–VK4)

 

 AC class   Availability (year)   Downtime (year)   Redundancy model   Typical application 

AC 1

99,0 %

 Up to 88 hours 

 No redundancy, single supply 

 Small businesses, test environments, laboratories 

AC 2

 99,9% 

 Up to 9 hours 

 Partial redundancy (N+1), one supply path 

 SMEs, internal IT, medium availability requirements 

 AC 3 

 99,99% 

 up to 53 minutes 

Full redundancy, multiple paths, maintainable

Enterprise IT, cloud, critical business processes

 AC 4 

 99,999 % 

Up to 6 minutes

 2N or 2(N+1), fault-tolerant 

 Financial sector, healthcare, KRITIS 

Note: While the Tier model defines specific downtimes, EN 50600 is based purely on the design architecture: The availability values are therefore for guidance only and may vary depending on implementation.

 

Availability targets: What different industries really need

Not every industry needs VK4 or Tier IV, but every industry needs clarity. Availability classes in data centres are a control tool for investments, risk assessments and service level agreements (SLAs). They help to align business models with infrastructure requirements. After all, the risks associated with a system failure vary dramatically. One hour of downtime can cost tens of thousands of dollars in e-commerce or lead to life-threatening situations in healthcare. The choice of availability class is therefore highly differentiated.

Critical infrastructures: the backbone of supply security 

Energy suppliers, waterworks, telecommunications providers: they operate facilities that are essential to the community. The requirements for data centre availability are correspondingly high. Systems must not only be redundant, but also fault-tolerant. A failure of such systems affects not only a single business, but entire regions, with consequences for security, supply, and public order.

  • Failures jeopardise security of supply and public order
  • VK4 or Tier IV is the minimum requirement in many cases – partly because national security standards and EU regulations require it
  • Redundancy, fault tolerance, and tested security architecture are essential

 

Healthcare: Data centres as lifelines 

In clinics, laboratories and emergency services, data centres secure medical technology systems, imaging procedures, patient records and surgical planning.

  • Even minutes without access can be critical
  • VK3 / VK4 or Tier III / IV levels are indispensable
  • High requirements also apply to physical security, redundancy and emergency management



Industry: Production reliability requires digital stability 

Automated production lines, supply chain control and machine communication rely on real-time data.

  • Failures lead to downtime, rejects and delivery delays
  • VK2 / Tier II may be sufficient – but VK3 / Tier III is increasingly becoming the standard
  • Availability in data centres is becoming a competitive factor in manufacturing



Public administration: Stability for government infrastructure 

Government agencies and municipal services process highly sensitive citizen data, operate e-files, procurement platforms and internal systems.

  • VK3 / VK4 or Tier III / IV – depending on the criticality of the service
  • Special requirements for data protection and access security
  • Certifications often a prerequisite for tenders

 

E-commerce: Real-time sales and real-time losses 

Online platforms operate around the clock. Every second counts and every downtime is measurable.

  • Studies show: up to 11,000 US dollars in lost revenue per minute of downtime
  • VK3 / Tier III are minimum standards for large shop systems
  • VK4 / Tier IV for platform operators with real-time transactions and payment systems

 

Financial service providers: availability as a compliance criterion 

Banks, insurance companies, stock exchanges and payment service providers are legally obliged to minimise IT failures.

  • VK4 / Tier IV are the de facto industry standard
  • Even short-term failures can lead to a loss of reputation and trust
  • Certifications are mandatory and required by law

 

The right class as a strategic decision 

Not every company needs to achieve VK4 or Tier IV. The decisive factor is the type of damage that could occur in the event of a failure: economic, legal or reputational. Choosing the availability class in the data centre helps to make these risks tangible and to negotiate appropriate SLAs with service providers.

Tip for decision-makers:
Don't automatically choose the highest class, choose the right one. Oversizing ( ) is expensive. Undersizing is risky.

 

How infrastructure affects the availability of a data centre

Availability is not achieved through individual measures, but through the interaction of all infrastructure elements – from power supply and cooling to operational organisation. Anyone who wants to achieve Tier III, Tier IV or VK3, VK4 must plan reliability systematically and holistically. The focus is always on the question: How is redundancy achieved and how fault-tolerant is the overall system?

Redundancy models: N, N+1, 2N, 2N+1

The decisive variable is the level of redundancy. Four models have become established internationally:

  • N – Basic configuration: Exactly the capacity required for normal operation – without reserves.
  • N+1 – One additional system per critical function (e.g. power, cooling, pumps). In the event of a failure, operation remains stable.
  • 2N – Two completely independent systems. Each can carry the full load on its own. Complete mirroring – everything remains active in the event of a fault or maintenance.
  • 2N+1 – Maximum fault tolerance: Two complete systems plus one additional backup each – for locations with absolute availability requirements.

These concepts define the technical basis for the respective Tier or VK level:

  • Tier II / VK2 relies on N+1
  • Tier III / VK3 requires N+1 with uninterrupted maintenance
  • Tier IV / VK4 requires 2N or 2N+1 – i.e. complete fault tolerance


Power supply: the foundation of high availability

High availability is not possible without a redundant power supply. In VK3/Tier III systems, power circuits are duplicated – with at least N+1 protection. Tier IV/VK4 requires completely separate power lines and UPS chains according to the 2N principle. Important: Redundancy only works if it is completely decoupled in terms of control technology. A supposedly dual supply without separation in the automation chain does not provide resilience.

 

Cooling: the thermal bottleneck

IT systems are sensitive to temperature fluctuations, which is why cooling is one of the most critical infrastructure elements. High-availability data centres rely on:

  • Redundant climate control cabinets (N+1 or 2N),
  • Separate recooling units
  • Automated switching between supply routes
  • Free cooling as an efficiency reserve

In VK4/Tier IV environments, the following applies: The failure of a chiller or cooling unit must not lead to a rise in temperature. Here too, complete fault tolerance must be guaranteed – both technically and operationally.

 

Security, network, fire protection: defence in depth

Modern data centres protect their function with multi-layered security concepts – analogous to the "onion skin principle":

  • Biometric access controls
  • Redundant carrier infrastructure with physical separation
  • Automatic early fire detection and inert gas extinguishing

The higher the selected tier or VK level, the more detailed the security and emergency concepts must be documented – including tests, training and documentation.

 

Component selection: the silent availability factor

Redundancy alone is not enough. The components used must themselves be fail-safe. An example: piping systems in cooling circuits. Faults here are rare, but have serious consequences: leaks lead directly to total failure. Component selection and material quality have a direct impact on the risk of failure, maintenance requirements and operational stability.

 High-quality pipe systems made of polypropylene (PP) such as aquatherm blue offer decisive advantages over metal pipes:

  • Corrosion resistance
  • Long service life (> 50 years)
  • Low weight
  • Easy installation
  • Low-maintenance installations
  • Fully recyclable

They are established worldwide – in Tier III / IV and VK3 / 4 certified facilities as well as in scalable cloud data centres. aquatherm blue was developed specifically for closed cooling and heating systems, is designed for high operational reliability and meets the requirements for long-lasting, reliable infrastructure in data centres. The homogeneous connection technology (fusion welding) ensures permanently tight connections – without seals, adhesives or detachable connections that could represent potential weak points. All in all, the more critical the data centre, the more the quality of the infrastructure determines long-term success.

 

Conclusion: Availability classes in the data centre as a strategic planning basis

Availability classes are a strategic basis for decision-making. They help to systematically evaluate IT infrastructure, secure planning and align operational concepts in a resilient manner.

There is no one-size-fits-all solution; a tailor-made architecture is crucial here. Not every data centre has to achieve VK4 or Tier IV. The class selection should be based on the business model:

  • How critical is an outage for our business operations?
  • What regulatory requirements exist in our industry?
  • What budgets and construction options are available to us?

The availability class in the data centre determines all subsequent decisions: from the floor plan and the choice of energy supply to the design of the cooling technology.

 Don't just plan a data centre. Plan availability.

aquatherm supports operators, planners and general contractors in the implementation of high-availability cooling infrastructures: with durable piping systems, consulting in early project phases and components that are designed to last for decades. Our products are used worldwide in data centres that have to operate around the clock.

Whether N+1, 2N or 2N+1: our systems meet the requirements of modern redundancy concepts and help to reliably achieve availability targets in accordance with EN 50600 and the Tier model: economical, resilient and future-proof.

We provide the technical foundation for global IT infrastructures in 24/7 operation. 

Talk to our experts.
aquatherm supports you from design and component selection to commissioning with industry-specific expertise and international project experience. 

Contact us