EXIN CDCS cheat sheet

Availability tiers (Uptime Institute): Tier I = single path, no redundant components, ~99.671% availability; Tier II = redundant components (N+1), ~99.741%; Tier III = concurrently maintainable, dual paths but only one active, ~99.982%; Tier IV = fault tolerant, two active paths, ~99.995%.

Tier III permits planned maintenance without shutdown; Tier IV additionally tolerates an unplanned failure of any single component. Concurrently maintainable and fault tolerant are the key discriminators on exam items.

Availability formula: Availability = MTBF / (MTBF + MTTR). Raising MTBF (reliability) or lowering MTTR (repair time) both increase availability; exam questions often ask which design change raises availability most.

Power path: utility feed → MV transformer → LV switchgear → UPS → PDU → rack PDU. Redundancy notations: N (no spare), N+1 (one spare), 2N (two full systems), 2(N+1) (two full systems each with a spare).

UPS topologies: offline/standby (load on raw mains, transfer on failure), line-interactive (AVR plus transfer), double-conversion online (load always on inverter, best power quality). Eco mode trades conditioning for efficiency.

Cooling: sensible heat ratio, supply/return air temperatures, and delta-T drive capacity. ASHRAE recommended inlet envelope is roughly 18–27 °C; raising inlet temperature within the envelope reduces cooling energy.

Airflow management: hot/cold aisle containment, blanking panels, and raised-floor perforations reduce bypass and recirculation; measure with return-air temperature and RCI (Rack Cooling Index).

PUE = Total facility power / IT equipment power. PUE is a ratio ≥ 1.0; it measures efficiency, not availability, and lower is better. DCiE = 1 / PUE.

Resilience vs. redundancy: redundancy duplicates components; resilience is the system's ability to absorb and recover from failure. Fault tolerance requires no interruption of load on a single failure.

Change and capacity management: baseline CMDB, approved change with rollback plan, and capacity thresholds (power, cooling, space, network) monitored against headroom targets.

Key standards: ISO/IEC 27001 (ISMS), ISO/IEC 22237 (data centre facilities and infrastructures), EN 50600 (European data centre design/operation), ANSI/TIA-942 (telecommunications infrastructure), ISO 50001 (energy management).

ISO/IEC 22237 and EN 50600 use 'availability classes' 1–4 and 'protection classes' rather than the Uptime Tier names; do not conflate the two schemes on exam items.

Risk treatment options: avoid, mitigate/reduce, transfer (insurance), accept. Residual risk remains after treatment and must be documented and owned.

Business Impact Analysis (BIA) determines RTO (recovery time objective) and RPO (recovery point objective); RTO drives redundancy design, RPO drives backup/replication frequency.

Security layers: physical perimeter, building envelope, secure zones per EN 50600 (e.g., computer room, secure technical space), rack, and logical access. Defence in depth uses layered controls.

Access control: least privilege, need-to-know, two-person rule for sensitive areas, visitor escorts, and audit trails. Biometrics plus badge plus PIN is multi-factor.

Fire suppression: clean agents (e.g., inert gas, FK-5-1-12) suit occupied IT spaces; water-based systems risk equipment damage. Detection uses VESDA/ASD early smoke detection.

Service management (ITIL-aligned): incident restores service fast; problem finds root cause; change is controlled; capacity and availability management are ongoing processes.

Continuity: DR site types (cold, warm, hot) map to RTO. A hot site with synchronous replication gives near-zero RPO but highest cost.

Compliance evidence: maintain logs, test records (UPS/generator load tests, failover drills), and audit findings; unverified redundancy is treated as unproven.

Monitoring stack: BMS/EPMS for power and cooling, DCIM for asset/capacity/power-chain mapping, network monitoring, and environmental sensors (temperature, humidity, leak, smoke).

Alarm philosophy: define severity levels, escalation paths, and deadbands to avoid alarm floods; every critical alarm should have a documented response procedure.

Preventive maintenance cadence: UPS battery impedance/load tests, generator exercise under load, cooling filter and belt checks, and infrared thermography on electrical connections.

Electrical safety: arc flash risk assessment, lock-out/tag-out (LOTO) before work, PPE per incident energy, and single-line diagrams kept current.

Cable management: separate power and data pathways, structured labelling, and documented patching to preserve traceability and airflow.

Efficiency metrics beyond PUE: WUE (water usage effectiveness), CUE (carbon usage effectiveness), and energy reuse effectiveness (ERE, where ERE = 1 − reuse/PUE-derived).

Free cooling/economizers: air-side or water-side economizers cut compressor run hours; effectiveness depends on climate and allowable inlet envelope.

High-density racks: liquid cooling (direct-to-chip, immersion) becomes relevant above roughly 15–20 kW per rack where air cooling struggles.

Decommissioning: data sanitization (NIST SP 800-88 categories clear/purge/destroy), asset record closure, and secure media destruction with certificates.

SLA management: define availability measurement window, exclusions, credits, and reporting; measure against agreed targets, not internal aspirations.