EPI Data Centre Associate cheat sheet

Availability tiers (Uptime Institute style): Tier I = single path, no redundant components, ~99.671% annual availability; Tier II = redundant capacity components, ~99.741%; Tier III = concurrently maintainable, ~99.982%; Tier IV = fault tolerant, ~99.995%.

Tier III requires every capacity component and distribution path to be serviceable without shutting down the IT load; Tier IV additionally requires a single worst-case fault to cause no IT outage.

Redundancy notation: N = load requirement; N+1 = one spare component; 2N = two full capacity sets; 2(N+1) = two full sets each with one spare.

Power usage effectiveness PUE = total facility energy / IT equipment energy; ideal 1.0, and DCiE = 1/PUE expressed as a percentage.

Airflow design rule: contain either the hot aisle or cold aisle, not both; supply air temperature commonly 18-27 C per ASHRAE TC 9.9 recommended envelope.

ASHRAE A1 allowable envelope: 18-27 C recommended; allowable range 15-32 C for A1 with humidity 8-90% RH (dew point limits apply).

Rack power density classes: <5 kW/rack air-cooled standard; 5-15 kW needs containment and careful airflow; >15-20 kW often requires liquid or rear-door heat exchangers.

Electrical chain efficiency: UPS losses typically 3-6%, PDU transformer losses 1-3%; each conversion stage subtracts from overall efficiency.

Cable pathway separation: power and data in separate trays or with a physical barrier; minimum separation commonly 300 mm where parallel runs are unavoidable.

Key standards: ANSI/TIA-942 (telecommunications infrastructure), ISO/IEC 22237 (facility and infrastructure), EN 50600 (European data centre series), BICSI 002 (design and implementation).

TIA-942 ratings are Rated-1 to Rated-4; the four levels map conceptually to Uptime Tiers I-IV but are separate certification schemes and are not interchangeable.

ISO/IEC 27001 covers the information security management system; ISO/IEC 27002 provides the control guidance; SOC 2 reports on trust services criteria (security, availability, confidentiality).

PCI DSS requires cardholder data environments to be physically secured with access logging and retention of at least 12 months of audit logs, with 3 months immediately available.

EN 50600-2-x covers building construction, power distribution, environmental control and telecommunications cabling; EN 50600-3-x covers operation.

Fire suppression in white space typically uses clean agents (FM-200, Novec 1230, inert gas) or water mist; sprinkler systems are permitted but risk water damage to IT.

Compliance evidence for audits: access logs, CCTV retention (often 90 days minimum), change records, UPS and generator test logs, and environmental alarm history.

Certification vs. compliance: a certificate (e.g., ISO 27001) is issued by an accredited body after audit; compliance with TIA-942 or EN 50600 may be self-declared or third-party rated.

Change management: every change to production infrastructure requires a request, risk assessment, approval, implementation plan, rollback plan, and post-implementation review.

MOP (method of procedure) is the step-by-step script for a specific task; SOP covers routine recurring operations; EOP covers emergency response.

Preventive maintenance intervals: generators load-tested monthly (no-load or exercise) and annually under load; UPS batteries tested annually with impedance checks; IR thermography on electrical panels annually.

Generator testing: monthly exercise run of at least 30 minutes; annual full-load bank test commonly 2-4 hours to verify capacity and fuel consumption.

Battery monitoring: VRLA design life typically 3-5 years; replace when capacity falls below 80% of rated or impedance rises significantly.

Environmental thresholds: cold aisle 18-27 C, humidity 40-60% RH recommended for ESD and corrosion control; alarm on high temperature, water detection, and smoke.

Incident severity: P1 = total outage or major customer impact; P2 = degraded redundancy; P3 = single component failure with redundancy intact; P4 = cosmetic or documentation.

Access control: two-person rule for critical areas, mantrap/vestibule entry, badge plus biometric, and visitor escort at all times with logged entry and exit.

Risk treatment options: avoid, mitigate, transfer (insurance), accept; residual risk must be documented and signed off by the risk owner.

Risk register entry fields: risk ID, description, likelihood, impact, inherent risk score, controls, residual score, owner, review date.

Business impact analysis produces RTO (maximum tolerable downtime) and RPO (maximum tolerable data loss); DR tiering follows from these values.

BCP/DR testing cadence: tabletop annually, functional test at least annually, full failover test per business requirement (often every 1-2 years).

Governance structure: an IT steering or data centre governance board with defined roles (facility owner, IT operations, security, compliance) and documented RACI.

Vendor management: SLAs must define availability percentage, measurement window, exclusions, credits, and reporting; review quarterly.

Availability math: 99.9% = 8.76 hours downtime/year; 99.99% = 52.6 minutes/year; 99.999% = 5.26 minutes/year.

Audit and assurance: internal audits at planned intervals, management review at least annually, and corrective actions tracked to closure in a register.