ISO/IEC 22237 & EN 50600 Data Centre Design Standard

Executive Summary

The data centre industry uses the word “standard” loosely. The Uptime Institute Tier system is a proprietary rating scheme; ANSI/TIA-942 is a telecommunications-led design standard; ASHRAE supplies thermal guidelines and an energy standard. Sitting across all of them is a formal, internationally governed design standard for data centre facilities and infrastructures: ISO/IEC 22237, the international counterpart to Europe’s EN 50600, paired with the ISO/IEC 30134 series that standardises the efficiency and sustainability KPIs. This trio is the closest thing the sector has to a single, holistic, vendor-neutral design-and-operate framework.

EN 50600 was developed by CENELEC and published in parts from around 2012. ISO/IEC then adopted the EN 50600 content — initially as Technical Specifications, progressively converted into full International Standards — as the ISO/IEC 22237 series under ISO/IEC JTC 1/SC 39. The two are deliberately aligned: ISO/IEC 22237 elevates EN 50600 to an international level so that data centres can be designed, built, and operated to the same principles worldwide. The companion ISO/IEC 30134 KPI series (and its EN 50600-4 European mirror) defines PUE, Renewable Energy Factor, Energy Reuse Factor, Carbon Usage Effectiveness, Water Usage Effectiveness, and the IT-equipment efficiency and utilisation metrics — so that a quoted “PUE” means one specific, audited thing rather than a marketing figure.

What the Standard Is

ISO/IEC 22237, formally Information technology — Data centre facilities and infrastructures, is a multi-part standard that specifies requirements and recommendations for the design, planning, procurement, integration, installation, operation, and maintenance of the physical infrastructure inside a data centre. It addresses the building itself, the power distribution chain, the environmental-control (cooling) systems, the telecommunications cabling, the physical security systems, and the management and operational information needed to run the facility against its defined objectives.

The standard is explicitly holistic. Rather than treating power, cooling, and security as separate disciplines, it organises them under a common framework of availability, physical security, and energy-efficiency objectives, each of which is derived from a business-risk analysis and then expressed as a class. A facility does not “pass” ISO/IEC 22237 in the abstract; it is designed and assessed to a chosen availability class and chosen protection classes, with the efficiency KPIs measured and reported against the ISO/IEC 30134 definitions.

It serves a broad audience by design: owners, operators, facility and ICT managers, and project managers on one side; consultants, architects, building and system designers, auditors, and test-and-commissioning agents on the other. Because it covers the full design-build-operate lifecycle, it is frequently referenced in tenders and procurement — a colocation customer or public buyer can specify “ISO/IEC 22237 Availability Class 3” as a contractual requirement and have it independently verified.

Why It Matters to Sustainability and Compliance Teams

For a long time this standard family was the concern of facilities engineers, not sustainability teams. That has changed as data centre energy and water use moved onto the regulatory agenda. Four pressure points now make it directly relevant to corporate sustainability and compliance work.

EU regulatory reporting. The EU Energy Efficiency Directive introduced a data-centre reporting and rating framework that draws on standardised infrastructure metrics — the ISO/IEC 30134 KPIs are the natural vocabulary for it. Operators reporting energy performance need PUE, and increasingly CUE and WUE, measured to a defensible definition rather than a marketing one.

Scope 2 and the electricity question. A data centre’s overhead — everything that is not the IT load itself — is captured by PUE, and its carbon by CUE. Both feed directly into Scope 2 electricity emissions accounting and the choice of electricity emission factor. A sustainability team reconciling a colocation provider’s reported emissions needs to understand what PUE and CUE do and do not include.

CSRD and energy disclosure. Under CSRD / ESRS E1, energy consumption and energy intensity are reportable, and for digital-infrastructure-heavy businesses the data centre is the dominant line. Standardised KPIs make those disclosures comparable and auditable.

Energy management integration. The standard dovetails with ISO 50001 energy management and the broader ISO management-system family. A facility certified to ISO/IEC 22237 with an ISO 50001 energy-management system in place has a coherent, auditable story from design intent through to continuous energy-performance improvement.

The Standard Family and Publication History

Three related document families are easy to confuse and worth separating cleanly: EN 50600 (the European design standard), ISO/IEC 22237 (its international counterpart), and ISO/IEC 30134 (the KPI series, mirrored in Europe by EN 50600-4).

The ISO/IEC 22237 Part Structure

The standard is organised into seven parts, each covering one infrastructure discipline. Part 1 sets the general concepts — including the availability and protection class framework — that the discipline-specific parts then apply.

The split matters in practice because availability is assessed per infrastructure. A facility’s power distribution, environmental control, and cabling can each carry a different availability class — a site might be Availability Class 3 for power but Class 2 for cooling, depending on how it was designed and where the business risk was judged to lie. There is no single facility-wide “Class” number that overrides the per-infrastructure assessment.

Availability Classes (AC 1–4)

The availability framework is the backbone of the design standard. Each of the relevant infrastructures — power distribution, environmental control, and telecommunications cabling — is assigned one of four availability classes, chosen from a business-risk analysis. The classes escalate from single-path, single-component up to fault-tolerant.

The escalation logic is two-dimensional: AC 1 → AC 2 adds redundant capacity components; AC 2 → AC 3 adds independent distribution paths so maintenance can occur without downtime (concurrent maintainability); AC 3 → AC 4 makes those paths active and fault-tolerant so a failure does not interrupt service. Higher classes cost more in capital, consume more in standby and conversion losses, and therefore carry a real sustainability trade-off — over-specifying availability is also over-specifying energy and embodied carbon.

Availability Class vs Uptime Tier vs TIA-942

The single most common confusion in data centre procurement is treating ISO/IEC 22237 Availability Classes, Uptime Institute Tiers, and ANSI/TIA-942 Ratings as interchangeable. They map roughly onto the same four-level resilience ladder, but they are governed differently, assessed differently, and are analogous, not equivalent.

The differences that matter:

• Assessment basis. ISO/IEC 22237 / EN 50600 assesses design architecture through planning, documentation, and infrastructure concept, within a holistic framework that also covers energy efficiency, security, and operations. The Uptime Tier system focuses on specifically verifiable fault tolerance and is a de-facto commercial standard, particularly in North America and Asia, with thousands of certificates issued worldwide.
• Governance. ISO/IEC 22237 and EN 50600 are formal, committee-governed standards (ISO/IEC JTC 1/SC 39; CENELEC). The Uptime Tier system is a proprietary scheme operated by a single body. TIA-942 is a telecommunications-industry standard.
• Uptime percentages. The Uptime Tier system is sometimes quoted with specific availability percentages (for example, Tier III at roughly 99.982% and Tier IV at roughly 99.995% annual availability). These figures belong to the Uptime framework and should not be assumed to attach identically to an ISO/IEC 22237 Availability Class — the standards do not define equivalence at that level of precision.

Protection Classes

Alongside availability, the standard defines a parallel framework of physical-security protection classes. Each data centre space — regardless of size or purpose — is designated to one of four protection classes, and the required class is identified separately for each physical-security objective.

As with availability, protection classes are chosen from a risk analysis rather than maximised by default — and they are space-specific, so a meet-me room, a data hall, and a plant room can carry different protection classes within the same facility. The four-class structure means a buyer can specify, and a certifier can verify, a precise physical-security posture rather than a vague “secure” assertion.

Power Distribution (Part 3)

Part 3 governs the electrical supply and distribution chain — from the utility intake through transformation, uninterruptible power supply, generation backup, and distribution to the ICT load. The availability class is expressed here through the redundancy of capacity components (transformers, UPS, generators) and the independence of distribution paths.

The power chain is where the availability/efficiency trade-off is sharpest. Redundant UPS and conversion stages held at part load to provide concurrent maintainability or fault tolerance carry conversion and standby losses, which show up directly in PUE. The standard also recognises the interaction between data centres and the wider grid — on-site energy storage and energy export change the resource-metric picture, and the KPI definitions account for how stored and exported energy is treated. A facility offering demand-response or exporting energy is not simply a load; its reported efficiency metrics have to handle those flows correctly.

Environmental Control (Part 4)

Part 4 covers the cooling and environmental-conditioning systems that keep IT equipment within its operating envelope. Cooling is typically the largest single component of data-centre overhead — the dominant driver of PUE above 1.0 — so it is the primary lever for both efficiency and water use.

The standard’s environmental-control requirements sit alongside the widely used ASHRAE thermal guidelines, which define recommended and allowable temperature and humidity envelopes for IT equipment. Designing to the wider end of the allowable envelope enables free-cooling (economiser) operation for more hours of the year, cutting energy use — but can increase water consumption where evaporative cooling is used, which is the tension that the Water Usage Effectiveness KPI (see §14) exists to expose. The arrival of high-density AI compute is pushing environmental control toward liquid cooling, which changes the efficiency and water calculus again and is an active area of standards development.

Cabling, Security, and Management (Parts 5–7)

Part 5 — Telecommunications cabling infrastructure

Part 5 specifies the structured cabling that carries data within the facility, aligned to the broader generic-cabling standards but adapted to the data-centre environment and its availability classes. At publish it remains a Technical Specification pending conversion to a full International Standard.

Part 6 — Security systems

Part 6 details the physical-security systems that implement the protection classes from Part 1 — access control, intrusion detection, fire detection and suppression, and environmental-hazard protection. It is one of the parts published as a full International Standard.

Part 7 — Management and operational information

Part 7 addresses the operational data and management processes needed to monitor a facility against its defined availability, security, and efficiency objectives. This is the part that connects the physical design to ongoing operation and to the KPI measurement that feeds regulatory reporting. At publish it remains a Technical Specification.

The Efficiency and Sustainability KPIs — ISO/IEC 30134

The ISO/IEC 30134 series is where the sustainability substance of the data-centre standard family lives. It defines the key performance indicators that quantify a facility’s energy efficiency, renewable and reuse performance, carbon, and water — each with a standardised calculation method so that a reported figure means one specific, comparable thing.

PUE in Depth

PUE is the ratio of total facility energy to IT equipment energy:

PUE = total facility energy ÷ IT equipment energy

A theoretically perfect facility, where every unit of energy reached the IT load and nothing was lost to cooling, power conversion, or lighting, would have a PUE of 1.0. Real facilities sit above that: the excess over 1.0 is the overhead — predominantly cooling, then power-conversion losses, then ancillary loads. A PUE of 1.5 means that for every unit of energy doing useful IT work, half a unit again is consumed by overhead.

PUE’s limits are as important as its definition. It measures infrastructure overhead, not useful work: a facility full of idle servers can post an excellent PUE while wasting energy, which is why ITEUsv (utilisation) is a necessary companion metric. It says nothing about the carbon intensity of the electricity (that is CUE) or about water (WUE). And it is sensitive to measurement boundary and period — an annualised, measured PUE to the ISO/IEC 30134-2 boundary is a very different number from a design-day estimate. Treat any single PUE figure as one facet of a multi-KPI picture, not a verdict.

Carbon (CUE) and Water (WUE) Effectiveness

PUE answers “how efficiently does this facility use energy?” but not “how clean is that energy?” or “how much water does it cost?” The two newer KPIs close those gaps.

Carbon Usage Effectiveness (CUE), defined in ISO/IEC 30134-8, expresses the carbon emissions associated with the data centre’s energy relative to its IT energy. Where PUE multiplied by the grid emission factor gives the carbon picture implicitly, CUE makes it an explicit, reportable KPI. A facility can have a strong PUE and a poor CUE if it sits on a carbon-intensive grid — which is exactly why location and Scope 2 electricity methodology, including the choice between location-based and market-based accounting, matter as much as physical efficiency.

Water Usage Effectiveness (WUE), defined in ISO/IEC 30134-9, expresses water consumed per unit of IT energy. Water became a first-order data-centre issue as evaporative and adiabatic cooling traded electricity for water in hot, often water-stressed, locations. WUE exposes the trade-off that a PUE-only view hides: a design that improves PUE by leaning on evaporative cooling can worsen WUE substantially. For facilities in water-scarce regions, WUE is increasingly the binding sustainability constraint, not energy.

Energy Reuse (ERF) and Heat Recovery

The Energy Reuse Factor (ERF), defined in ISO/IEC 30134-6, is the ratio of energy reused to total energy consumed in the data centre. It captures the recovery and productive reuse of energy that would otherwise be rejected as waste heat — most prominently, supplying recovered heat to a district-heating network or to neighbouring buildings.

Heat recovery is the most structurally interesting frontier in data-centre sustainability because it reframes the facility from a pure consumer into a node in a wider energy system. A data centre rejects large quantities of low-grade heat; where that heat can be captured and delivered to a heat network, the ERF rises and the net system-level efficiency improves even if the facility’s own PUE is unchanged. Note the important interaction: PUE and ERF measure different things and can move independently — a facility can have an unremarkable PUE and still deliver substantial value through a high ERF. Reporting both is what tells the full story.

Conformity and Certification

Conformity to ISO/IEC 22237 (and EN 50600) is demonstrated through independent assessment, and there are two distinct routes that practitioners frequently conflate.

The distinction is consequential in procurement: a design certificate confirms intent, not delivery. A buyer specifying ISO/IEC 22237 conformity should be clear whether a design or a facility certificate is required — a design certificate on a facility that was then built differently does not demonstrate operational conformity. Certification is performed by accredited certification bodies; several offer combined assessment against TSI (Trusted Site Infrastructure), EN 50600, and ISO/IEC 22237 in a single engagement.

The standard also complements the ISO management-system family rather than competing with it. ISO/IEC 22237 addresses the quality and availability of the physical infrastructure; ISO 50001 addresses the energy-management system that drives continuous efficiency improvement; ISO/IEC 27001 addresses information security; and organisation-level GHG quantification under ISO 14064-1 sits above all of them. A mature operator runs these together — infrastructure conformity, energy management, security management, and carbon accounting as one coherent assurance stack.

Interaction with EU Regulation

The standard family’s regulatory weight comes chiefly from the EU. The recast Energy Efficiency Directive introduced obligations around data-centre energy and sustainability reporting, and a Union-level scheme for rating and disclosing data-centre sustainability indicators. The ISO/IEC 30134 KPIs are the natural metric vocabulary for that reporting — PUE, REF, ERF, CUE, and WUE are exactly the indicators a rating scheme needs, defined to a common, auditable method.

Two consequences follow for corporate teams. First, operators in scope of the EED data-centre provisions need their KPIs measured to the standardised definitions, not to marketing conventions — a non-standard PUE will not survive a reporting obligation that references the standardised metric. Second, the energy and intensity datapoints these metrics feed flow upward into broader disclosure under CSRD / ESRS E1, and the electricity totals feed Scope 2 accounting via the relevant grid emission factors. The design standard, the KPI series, and the disclosure regime form a connected chain from facility design to corporate climate reporting.

Interaction with Other Standards

ISO/IEC 22237 coexists with several other data-centre standards rather than replacing them. Understanding the division of labour prevents both duplicated effort and incorrect substitution.

The recurring trap is treating the resilience schemes (ISO/IEC 22237 AC, Uptime Tier, TIA-942 Rated) as one thing and the KPI/management standards as optional extras. In a mature programme they are layered: a resilience class for the infrastructure, the ISO/IEC 30134 KPIs for efficiency and sustainability, and the ISO management systems for ongoing assurance.

Common Misinterpretations

Common Implementation and Reporting Errors

1. Quoting design PUE as measured PUE. A design-stage estimate is not an operational, measured, annualised figure. State explicitly which you are reporting, the boundary, and the period.
2. Reporting PUE without a utilisation metric. A strong PUE on a fleet of idle servers hides wasted energy. Pair PUE with ITEUsv (or an equivalent utilisation measure).
3. Treating availability class as facility-wide. Assess and report power, cooling, and cabling classes separately; do not collapse them into one number.
4. Cross-walking certifications. Asserting Uptime Tier III “equals” Availability Class 3 in a contract or disclosure. They are analogous, not equivalent; each requires its own assessment.
5. Optimising PUE at the expense of water. Leaning on evaporative cooling to cut PUE can worsen WUE materially, especially in water-stressed regions. Report and manage both.
6. Ignoring CUE on a clean-PUE facility. A physically efficient facility on a carbon-intensive grid still has high emissions. Carbon depends on the grid factor and the Scope 2 method, not just on PUE.
7. Citing a superseded KPI edition. The ISO/IEC 30134-2 PUE definition was revised in 2026; figures computed to an older boundary may not be comparable. Cite the edition.

Implementation Workflow

Future Evolution

TS-to-IS conversion. The remaining ISO/IEC 22237 parts held as Technical Specifications (notably cabling and management/operational information) are on a path to becoming full International Standards. Practitioners should expect the published-IS coverage to broaden over the next standards cycles, with implementation lead times of a couple of years.

AI density and liquid cooling. The shift to high-density AI compute is straining the air-cooling assumptions behind much existing environmental-control design. Liquid cooling changes the efficiency and water calculus and is driving active development across the environmental-control and KPI parts of the family.

Tightening EU regulation. The EU’s data-centre reporting and rating framework under the Energy Efficiency Directive is expected to sharpen over the late 2020s, raising the stakes on standardised, audited KPIs and on water reporting in particular. The direction of travel — mandatory, comparable, standardised metrics — is firmly set.

New efficiency KPIs. Beyond the established ISO/IEC 30134 set, work continues on metrics that capture useful-work efficiency more directly — server energy effectiveness and application-platform energy effectiveness among them — reflecting the recognition that infrastructure-overhead metrics like PUE do not, on their own, measure whether compute is being delivered efficiently.

Frequently Asked Questions

Sources and References

Every methodological statement on this page reconciles to the primary sources below. Figures and part references are hardcoded against the dated primary sources.

Primary standards

• ISO/IEC 22237 series, Information technology — Data centre facilities and infrastructures, Parts 1–7 (ISO/IEC JTC 1/SC 39). Part 1 General concepts; Part 2 Building construction; Part 3 Power distribution; Part 4 Environmental control; Part 5 Telecommunications cabling infrastructure; Part 6 Security systems; Part 7 Management and operational information.
• EN 50600 series, Information technology — Data centre facilities and infrastructures (CENELEC), the European counterpart from which ISO/IEC 22237 derives, including the EN 50600-4 KPI sub-parts.
• ISO/IEC 30134 series, Information technology — Data centres — Key performance indicators: Part 2 Power Usage Effectiveness (PUE); Part 3 Renewable Energy Factor (REF); Part 4 IT Equipment Energy Efficiency for servers (ITEEsv); Part 5 IT Equipment Utilization for servers (ITEUsv); Part 6 Energy Reuse Factor (ERF); Part 7 Cooling Efficiency Ratio (CER); Part 8 Carbon Usage Effectiveness (CUE); Part 9 Water Usage Effectiveness (WUE). ISO/IEC 30134-2 revised 2026.

Related standards and schemes

• ANSI/TIA-942, Telecommunications Infrastructure Standard for Data Centers (Rated-1 to Rated-4).
• ASHRAE Standard 90.4, Energy Standard for Data Centers, and ASHRAE thermal guidelines for data processing environments.
• Uptime Institute Tier Classification System (Tier I–IV) — proprietary resilience rating.
• ISO 50001 (energy management), ISO/IEC 27001 (information security), ISO 14064-1 (organisation-level GHG quantification).
• Trusted Site Infrastructure (TSI) certification scheme.

Regulatory context (status as of review date)

• EU Energy Efficiency Directive (recast) — data-centre energy and sustainability reporting obligations and the Union data-centre rating scheme.
• Corporate Sustainability Reporting Directive (CSRD) and ESRS E1 (climate change) — energy consumption and intensity datapoints.

Note on the requested title

• This page was requested as the “SDIA Data Centre Design Standard.” No standard of that name exists. The Sustainable Digital Infrastructure Alliance (SDIA) published a Roadmap to Sustainable Digital Infrastructure and a Data Center Sustainability Integration Model and transitioned to an independent consultancy from 2025; it did not issue a formal data centre design standard. This reference therefore documents the actual data-centre design standard family (ISO/IEC 22237 / EN 50600 + ISO/IEC 30134).

What changed in this revision

Reviewed 19 June 2026. Initial publication. Documents the ISO/IEC 22237 / EN 50600 data centre facilities and infrastructures standard family and the companion ISO/IEC 30134 KPI series: the seven-part structure, the four availability classes, the four protection classes, the design-vs-facility certification routes, and the full efficiency/sustainability KPI suite (PUE, REF, ITEEsv, ITEUsv, ERF, CER, CUE, WUE). Covers the crosswalk to Uptime Tiers and TIA-942 (analogous, not equivalent), the interaction with the EU Energy Efficiency Directive data-centre reporting framework and CSRD / ESRS E1, and the integration with ISO 50001, ISO/IEC 27001, and ISO 14064-1.