EPI Data Centre Foundation key terms

The proportion of time a data centre system is capable of performing its intended function, usually expressed as a percentage over an agreed measurement period. It is calculated from uptime and downtime and underpins service commitments made to customers.

The provision of extra capacity or duplicate components so that a single failure does not stop the system working. Redundancy can be applied to power, cooling, networks or staff, and is usually described using N, N+1 or 2N notation.

The ability to carry out planned maintenance on any part of the infrastructure while the IT load continues to be served without interruption. It is a defining characteristic of higher-tier designs and requires redundant capacity plus documented procedures.

The capability of a system to keep operating, without loss of service, when one or more of its components fail. It usually relies on duplicated paths or components and automatic switching, and exceeds simple concurrent maintainability.

Any component, path or dependency whose failure would cause the whole system or a critical service to stop. Identifying and removing these is a core objective of resilience design and risk assessment.

A battery-backed device that supplies conditioned power to IT equipment for a short period during a mains failure or disturbance, bridging the gap until generators start or the load is shut down cleanly.

The ratio of total facility energy consumption to the energy delivered to IT equipment. A PUE of 1.0 would mean no overhead; lower values indicate greater energy efficiency and are widely used as a design and reporting metric.

The systems that remove heat generated by IT equipment and keep inlet temperatures within equipment limits. Approaches include room-based air conditioning, containment, in-row coolers and liquid cooling, and they strongly influence energy use.

A layout and physical barrier arrangement that separates cool supply air from hot exhaust air in the white space. It improves cooling efficiency and reduces mixing, allowing higher set points and lower energy consumption.

Industry guidance that defines recommended and allowable ranges for temperature and humidity at IT equipment inlets. It classifies environments into classes and is widely referenced when setting data centre operating envelopes.

An automatic device that transfers an electrical load from one power source to another within milliseconds when the preferred source fails or drifts out of tolerance. It is common in dual-corded distribution designs.

A physical network or fibre link between two points, often between customers or carriers within a data centre. Cross-connects are a core interconnection product and must be documented and controlled for security and capacity.

A controlled space where telecommunications carriers terminate their networks and interconnect with each other and with customers. Access is restricted and cross-connections are typically ordered and recorded through a formal process.

The structured routing, labelling and support of power and data cabling to maintain airflow, enable maintenance and prevent errors. Good practice includes separation of power and data and accurate records of every connection.

A controlled process for planning, approving, implementing and reviewing changes to infrastructure or services. It aims to reduce unintended outages by assessing risk, scheduling work and providing rollback plans before changes occur.

A pre-agreed period during which planned work, such as testing or repair, may take place with reduced or suspended service. It is defined in the service agreement and coordinated with the customer.

The discipline of recording and controlling the components, versions and relationships that make up the infrastructure and services. Accurate configuration data supports change planning, fault diagnosis, audits and capacity decisions.

The process of monitoring and planning power, cooling, space, network and staffing resources so demand can be met without overloading systems. It balances current utilisation against forecast growth and contractual commitments.

Software and sensors used to monitor and manage assets, power, cooling and space in a data centre. It provides a live model of the facility that supports capacity planning, efficiency reporting and incident response.

Layered controls that prevent unauthorised access to facilities and equipment, such as perimeter fencing, access control, CCTV, visitor escorts and locked cabinets. Controls should be proportionate to risk and regularly tested.

The systems and rules that determine who may enter which areas and when, using methods such as badges, biometrics, PINs and logging. Records support investigation and demonstrate compliance during audits.

The process for detecting, recording, responding to and recovering from unplanned events that affect service. It includes escalation, communication with customers and a post-incident review to prevent recurrence.

The practice of finding and eliminating the underlying causes of recurring incidents. It uses trend analysis and root cause investigation to produce permanent fixes, reducing repeat disruption and improving stability.

A structured method, such as the five-whys or fishbone diagram, used to identify why an incident happened rather than only what happened. Findings should feed corrective actions and lessons learned.

A contract between a provider and a customer that defines the service, performance targets such as availability and response times, and remedies if targets are missed. It should be measurable and unambiguous.

An independent, documented examination of evidence to determine whether controls, records and practices meet stated requirements. Data centre audits may cover security, energy, operational processes or contractual compliance.

The state of meeting applicable laws, regulations, standards and contractual obligations. It requires identifying requirements, implementing controls, keeping evidence and reviewing performance through internal checks and external audits.

Documented or observable proof that a control is designed and operating as intended, such as logs, test records, training certificates or policies. Auditors rely on evidence rather than verbal assurance to reach conclusions.

A published, consensus-based document that sets requirements, specifications or good practice, often maintained by a recognised body. Standards may be certifiable or guidance only, and adoption is usually voluntary unless referenced by contract or regulation.

Formal confirmation by an accredited body that an organisation, facility or individual meets a specified standard. It usually involves assessment against criteria and periodic surveillance to maintain validity.

A management process that identifies critical activities and builds resilience so essential services can continue or recover during disruption. It covers business impact analysis, recovery plans, testing and regular review.

The plans, facilities and technologies used to restore IT services and data after a major disruption. Recovery objectives for time and data loss are agreed in advance and validated through testing.

The maximum acceptable duration for restoring a service after disruption, agreed with the business. It drives the choice of standby sites, spare capacity and staffing needed to meet continuity commitments.

The process of identifying threats and vulnerabilities, estimating likelihood and impact, and prioritising treatment. Results guide controls, investment and acceptance decisions, and should be reviewed when conditions change.

A controlled record of identified risks, their owners, ratings and planned treatments. It provides a live view of exposure and is a key artefact for governance meetings and audits.

The framework of roles, policies, decision rights and oversight that directs how a data centre is managed and held accountable. It ensures objectives are set, risks are owned and performance is reported to senior management.

The combination of training, qualification, experience and demonstrated performance that enables a person to carry out a task safely and correctly. It should be defined for each role and verified through records.