Digital services are everywhere, but their physical foundation – the data centre – is vulnerable. As the infrastructural backbone of modern society, data centres are critical, yet the remain risk hotspots. International studies show that in 25% of cases, failures cost between $300,000 and $400.000 per hour – and such incidents are becoming increasingly frequent.
Power failures are the most common single risk in data centres. They cause 43% of all significant disruptions. An example from 2025 illustrates the extent of the problem: in Virginia, 60 data centres were simultaneously disconnected from the grid. The result was a regional supply crisis and proof that even clusters remain susceptible to failure if there is a lack of structural redundancy in data centres.
Thermal risks are also increasing. Just a few minutes without functioning cooling is enough to permanently damage server systems. Heat events – increasingly exacerbated by climate change – are pushing air conditioning systems to their limits. Cooling failure means more than just a rise in temperature: it directly jeopardises the integrity of the hardware.
Added to this are targeted cyber attacks. Modern data centres are not only the target of digital data theft, but also physical sabotage. Attackers are increasingly targeting building technology, networks, cooling systems, and power supplies with the aim of paralysing entire systems. The infrastructure itself becomes a target.
And finally, system failure due to internal errors – whether due to defective components, human error, or lack of maintenance. The failure of just one element can set off a chain reaction. Without adequate safeguards, this can quickly lead to total system shutdown.
In IT infrastructure, redundancy in a data centre refers to the deliberate duplication of critical systems. Its purpose is to eliminate so-called single points of failure – those points where the entire system could collapse if they fail. The core principle is simple: if one component fails, a second one takes over immediately – without interruption, data loss, or escalation. In environments where high availability is required, no single element should be critical on its own, whether it’s power, cooling, networking, early fire detection, or data processing. Every system is designed so that a fault is compensated for before it can have any effect, as is standard practice in a redundant data centre.
In practice, a distinction is made between two basic forms:
International standards set the direction:
The international Uptime Institute classifies data centres into four availability classes (tiers). These metrics help operators to classify their own risk and demonstrate it to customers, insurers, and auditors.
| Redundancy model | N+1 | 2N | 2N+1 |
|
Description |
Minimum number of components (N) + one additional backup component |
Complete duplication of critical infrastructure |
2N redundancy + one additional layer of security |
|
Example |
5 servers for normal operation → 6 servers in total |
200 kW demand → two independent 200 kW systems |
200 kW demand → two 200 kW systems + additional capacity |
|
Availability |
Up to 99.749% (Tier II) |
Up to 99.982% (Tier III) |
Up to 99.995% (Tier IV) |
|
Max. downtime/year |
22 hours |
1.6 hours |
26.3 minutes |
|
Cost |
Low - Cost-effective |
High – Dual infrastructure |
Very high - Highest investment |
|
Space requirements |
Low – Only one additional component |
High – Doubling of infrastructure |
Very high - Maximum space requirement |
|
Energy consumption |
Moderate – Only backup component |
High - Dual systems |
Very high - Maximum consumption |
|
Fault tolerance |
Single component failures |
Multiple simultaneous failures, complete system failure of a strand |
Maximum fault tolerance, multiple system failures |
|
Maintenance |
Maintenance possible during operation (limited) |
Complete maintenance possible during operation |
Maximum maintenance flexibility |
|
Area of application |
Small to medium-sized data centres |
Business-critical applications |
Mission-critical applications (healthcare, finance) |
A redundant data centre is not simply a duplicate server rack, but a comprehensively orchestrated system consisting of physical infrastructure, technical architecture, and operational logic. A redundant data centre must be thought through consistently across all levels.
Power
The power supply is at the heart of the system. It is not only essential, but also the most common cause of failure. That is why redundancy begins here with the connection to two independent external networks. Inside, uninterruptible power supply (UPS) systems bridge the gap between power failure and emergency power operation.
Air conditioning
Air conditioning must also be more than just efficient. It must be fail-safe, segmentable, and independently controllable. Modern systems combine cold and hot aisle principles with liquid-based cooling circuits. Water pipes with weldable plastic pipe systems, especially PP pipes, are becoming increasingly important in this context. They are corrosion-free, pressure-resistant, durable and require significantly less maintenance than metal alternatives.
Fire protection
When it comes to fire protection, it is not only active systems such as smoke detectors and gas extinguishing systems that are crucial, but also their integration into a redundant infrastructure. Depending on the tier level, fire compartments, sensors, and control units must be available in multiple instances and logically decoupled so that no single event compromises the entire security system.
Network connection
The network infrastructure forms the digital lifeline. Without a stable data connection, even the best internal redundancy is of little use. That is why high-performance data centres have at least two carrier connections, physically separate routes, and duplicate network nodes.
Media lines
Media routing is also crucial to long-term success. The choice of materials, routing, and tightness play a central role, especially in cooling and fire-fighting water systems. PP-based plastic piping systems have become the norm here because they are durable, flexible, and safe.
According to ITIC (Information Technology Intelligence Consulting), over 90% of companies now demand availability of at least 99.99%. In highly regulated or data-driven industries, over 40% even strive for 99.999% availability. This not only protects business processes, but also secures customer relationships, contract compliance, and market acceptance.
Protection against immense damage
According to ITIC, the average damage per hour of downtime exceeds one million US dollars in 41% of large companies. In critical time windows – such as stock market trading or medical systems – this figure increases per minute. Against this backdrop, redundancy is not a cost factor for data centres, but a risk buffer with a clearly calculable ROI.
Operations remain stable – even in the event of a malfunction
Redundant systems guarantee business continuity: applications continue to run, data remains accessible, and processes are not interrupted. Whether it's a cyberattack, power failure or software error, critical business processes are not interrupted. For many industries, this is not only economically relevant, but also a regulatory requirement.
Errors lose their impact
Whether it's a hardware defect, human error, or software problem, the real damage only occurs when systems fail to respond. Redundant architectures eliminate this escalation effect: errors are limited locally, systems switch over automatically, and data remains consistent.
Trust and legal certainty
Certified redundancy is a strategic advantage. Operators who meet Tier III or Tier IV levels and ISO/IEC 27001 show planners, customers, and regulatory authorities that they take availability seriously. This reduces regulatory risk, strengthens auditability, and contributes to brand trust.
The basis for fail-safe data centres is not technology, but planning. Redundancy can only be effectively implemented in the data centre if structural, logistical, and regulatory requirements are taken into account at an early stage. Many weaknesses do not arise during operation, but due to inadequate infrastructure concepts in the project planning phase.
Location factors: power, data, climate
A suitable location offers access to a stable power supply, ideally via several independent electricity providers. Proximity to medium or high-voltage grids is crucial from an economic perspective. Internet hubs such as DE-CIX (Frankfurt) or AMS-IX (Amsterdam) are relevant for grid connection, especially for latency-sensitive applications. In addition, the local climate influences operating costs: cooler regions significantly reduce energy requirements, as up to 40% of electricity consumption is accounted for by air conditioning.
Security aspects: risk minimisation through location selection
Data centres should be planned in areas with minimal risk of natural disasters. The BSI recommends minimum distances of 40 km from nuclear facilities and 1 km from petrol stations. Physical security criteria – such as low crime rates, controllable access zones, or the possibility of redundant transport and delivery routes – are also crucial. In addition, there are legal requirements: data protection laws and national sovereignty rules, such as in Germany, where from 2026 data centres may only be built near a potential heat consumer in order to use the waste heat as local heating.
UPS systems as risk zones
The uninterruptible power supply (UPS) is considered a critical component and, paradoxically, is the most common source of failure. According to evaluations, UPS malfunctions cause around 43% of all serious failures. The main causes are ageing batteries, control errors, overload, and inadequate maintenance. Effective protection therefore requires not only N+1 or 2N design, but also close maintenance schedules and redundant control mechanisms.
Maintenance as a system-critical process
Redundant systems only remain available if they are maintained. Preventive maintenance – i.e. the proactive replacement of critical components – is therefore essential. Air conditioning technology, fire alarm systems, extinguishing technology, monitoring systems, and low-voltage main distribution are particularly affected. Maintenance concepts should include remote monitoring in addition to on-site checks. Logging is crucial: only documented processes can be quickly analysed and improved after malfunctions.
High availability at the system level
Redundancy does not end with the physical infrastructure. High availability is only achieved through software mechanisms. These include failover logic, redundant applications, cluster architectures, and database mirroring. The goal is to minimise unplanned downtime – ideally to less than five minutes per year. In many industries, these values are not only best practice, but also a regulatory requirement.
Redundancy is the key stability factor for digital infrastructures. Today, redundant data centres support business-critical processes, supply industrial systems, and secure government functions. Their availability determines business continuity, compliance, and trust.
The analysis shows that redundancy minimises downtime in the data centre, protects against system failure, and enables business continuity – even under disruptive conditions. Holistic planning of location, infrastructure, operation, and maintenance reduces risks and increases the strategic value of IT. Standards such as ISO/IEC 27001 and Tier IV classifications make these requirements measurable.
Planners and operators should therefore view redundancy in the data centre not as a cost factor, but as a value driver. Early investment in dual supply paths, independent cooling circuits, and maintenance-friendly components not only protects against failures, but also safeguards reputation, efficiency, and regulatory compliance. Choosing the right system partners is crucial.
Talk to aquatherm about your project
aquatherm supports operators in the implementation of safety-relevant cooling infrastructures with corrosion-free, pressure-stable PP-based piping systems that have been specially developed for use in data centres. As an experienced solution provider, aquatherm accompanies projects from planning and design through to implementation.
Put your trust in a partner with comprehensive expertise in the cooling of state-of-the-art data centres and innovative pipe systems.
Contact our experts.
