Free CDCP Exam Flashcards

Basic Capacity site: single, non-redundant distribution paths for power and cooling. Any component failure or planned maintenance disrupts IT, so Tier I only suits non-critical workloads. Reference availability target is 99.671% (~28.8 h/year downtime).

Adds Redundant Capacity Components (N+1) to power and cooling, but keeps a single distribution path. A failed UPS module or chiller is covered; a fault on the shared feed is not. Reference availability is 99.741% (~22 h/year).

Multiple independent distribution paths let every capacity component and path be removed for planned maintenance without dropping IT load. Only one path must be active at a time. Reference availability is 99.982% (~1.6 h/year).

Goes beyond Tier III by tolerating any single unplanned failure, not just planned work. Requires 2N (or 2(N+1)) on all critical systems with compartmentation so one event cannot take down both sides. Reference availability is 99.995% (~0.4 h/year).

TIA-942 uses 'Rated' (not 'Tier' — a Uptime Institute trademark) for facility robustness. Rated-1 Basic, Rated-2 Redundant Components, Rated-3 Concurrently Maintainable, Rated-4 Fault Tolerant. They map conceptually to Uptime Tiers but are assessed against TIA's own criteria, so they are not interchangeable certifications.

European/global multi-part data centre standard covering design and operations. It classifies on Availability (1-4), Physical Security (1-4), and Energy Efficiency Enablement, plus KPIs (PUE, WUE, CUE per ISO 30134). EN 50600 is the EN version; ISO/IEC 22237 is its international counterpart.

N is baseline capacity sized exactly to the load — no spare, so any failure causes downtime. N+1 adds one redundant unit (e.g. one extra UPS module or chiller) so a single component failure is absorbed, but a single shared distribution path is still a single point of failure.

2N runs two fully independent, identical systems — either can carry 100% of load — eliminating single points of failure across the whole distribution chain (required for Tier IV). 2(N+1) goes further: each of the two systems has its own spare component, surviving one failure on each side at once.

TIA-942 and EN 50600 site assessments reject locations adjacent to refineries, chemical plants, fuel depots, rail lines, and flight paths because of explosion, toxic plume, or crash risk. Industrial standoff distance is a hard disqualifier, independent of seismic, flood, or hydrological scoring.

Modern high-density cabinets can exceed 1500 kg fully populated, so mission-critical raised floors are typically specified for >=1465 kg/m2 (300 lb/ft2) uniform load plus concentrated and rolling-load ratings. Lighter office-grade tiles (<=1000 kg/m2) risk point loading and fracture under rolling loads.

A dropped ceiling below the structural slab creates a return-air plenum, houses lighting and cable trays, and conceals fire-suppression piping. In hot-aisle containment designs the ceiling plenum often serves as the hot return path back to CRAH units.

Strong electromagnetic fields from high-voltage transmission lines, transformers, or rail systems can disrupt IT equipment and corrupt storage media. Site surveys measure DC and AC field strength; mitigation is shielding (mu-metal, ferromagnetic enclosures) or relocating the white space away from the source.

Conductive cable sensors snaked under the raised floor and around CRAC units, chillers, and pipe runs alarm before pooled water reaches live power or subfloors. Zoned detection tells operators which row or CRAC area is wet, enabling fast isolation and limiting water damage.

The load runs on raw utility power; the inverter only engages after an outage is detected. There is a 4-10 ms break that IT power supplies can ride through, but the load is exposed to sags, surges, and harmonics the whole time. Suitable for non-critical office IT, not data centre white space.

Adds an autotransformer/buck-boost to offline topology for voltage regulation without switching to battery, trimming minor sags and swells. The load still sees utility power quality during normal operation, so it is mid-tier protection — better than offline but not isolation-grade.

AC is continuously rectified to DC and inverted back to AC, isolating the load completely from utility disturbances. Zero transfer time and the cleanest output — standard for Tier III/IV. Efficiency runs 94-97% in normal mode; eco-mode trades some isolation for higher efficiency.

Typical VRLA or lithium-ion strings are sized for 5-15 minutes at full load — long enough for the diesel generator to start, synchronise, and accept load (usually 30-90 s). Longer runtimes grow battery cost and footprint linearly and are rarely justified where standby generators are present.

Senses loss of utility and transfers the load to the standby generator. Open-transition ATS breaks before making (typical <100 ms). Closed-transition or soft-load ATS overlaps sources to avoid the break. ATS handles coarse source selection; STS handles fast sub-cycle switching inside the room.

Thyristor-based switch that transfers between two synchronised AC sources in ~4-8 ms — faster than IT power supplies perceive an interruption. Used at PDU level in 2N Tier IV designs so dual-corded loads can draw from either feed without mechanical breaker operation.

Carries the data centre through extended utility outages after the UPS batteries ride through the start window. Sized to support full IT plus cooling and critical mechanical load (NFPA 110 Level 1 Type 10 typical). On-site fuel must cover the site's replenishment horizon, commonly 24-72 hours.

A bonded copper mesh installed under the raised floor ties all racks, cabinets, and equipment chassis to a common reference potential, shunting high-frequency noise and equalising ground voltage. It complements (does not replace) the life-safety electrical bonding network.

Non-linear loads (UPS rectifiers, VFDs, servers) inject harmonic currents that overheat neutrals, trip breakers, and reduce generator capacity. IEEE 519 sets recommended limits (e.g. <5% THD at the point of common coupling). Mitigation: passive harmonic filters, 12-pulse rectifiers, or active front-end UPS.

A single push-button or stop-chain action that cuts power to the white space to protect life during fire or electrocution events. NFPA 70 and 75 require it in IT rooms, but accidental EPO activation is a leading cause of data centre outages, so design must guard against misoperation (covers, dual-action, staged shutdown).

A CRAC (Computer Room Air Conditioner) contains a compressor and direct-expansion (DX) refrigerant coil — self-contained refrigeration. A CRAH (Computer Room Air Handler) has no compressor; it uses a chilled-water coil fed by a central chiller plant. CRAHs scale better in large facilities served by efficient chiller plants.

Cabinet rows are alternated so server intakes face each other across a cold aisle and exhausts face each other across a hot aisle. This stops hot exhaust from mixing back into the cold supply, raises return-air delta-T at the cooling unit, and improves both efficiency and inlet-temperature consistency.

Cold-aisle containment seals the cold intake aisle and lets the rest of the room run warm (return plenum) — staff working in the room feel the heat. Hot-aisle containment seals the hot exhaust aisle and keeps the wider room at cool supply temperature, so personnel stay comfortable. Cooling efficiency is similar; comfort and equipment accessibility differ.

Server inlet air should be 18-27 C (64.4-80.6 F) with 20-80% RH. This single recommended envelope is the SAME across classes A1-A4 — what differs between classes is the ALLOWABLE range you may temporarily exceed. Designing to the recommended envelope balances reliability and cooling energy.

A1: 15-32 C / 20-80% RH (tightest, enterprise). A2: 10-35 C. A3: 5-40 C. A4: 5-45 C / 8-90% RH (widest, for aggressive economizer use). Widening the allowable envelope lets facilities run higher supply temperatures and more free-cooling hours, but equipment warranty and reliability must support the class.

When outside air temperature and humidity fall within ASHRAE allowable envelopes, filtered outside air is brought directly into the white space and the compressors shut off — the biggest PUE reducer available in suitable climates. Requires high-efficiency filtration, humidification control, and management of dust or pollution events.

Cooling-tower water is run through a heat exchanger (plate-and-frame or in the chiller's condenser barrel) to produce chilled water without running the chiller compressor. Outside air never enters the white space — better for filtration and pollution concerns, but less efficient than air-side because of the extra heat-exchange step.

A cold plate with microchannels is bolted to CPU/GPU dies; a coolant loop (water/glycol or engineered fluid) carries heat directly away. It rejects far higher heat flux than air (kW per rack instead of kW per row) while leaving memory and power delivery air-cooled. The rest of the server remains in the airflow.

Servers are fully submerged in a dielectric fluid. Single-phase pumps the fluid through a heat exchanger; two-phase lets the fluid boil at chip temperature and condense at the condenser coil, transferring huge heat with no pumps in the boiling loop. PUE can drop below 1.05, but the server field is sealed and the fluid is expensive.

RDHx replaces the cabinet rear door with a chilled-water coil that absorbs exhaust before it enters the room — turning each rack into its own cooler. In-row units sit between racks and short-circuit supply air to intakes. Both approaches target high-density zones without raising room-wide cooling capacity.

NFPA 75 is the Standard for the Fire Protection of Information Technology Equipment — covering construction, suppression, and emergency procedures for IT rooms and data centres. NFPA 76 covers telecommunications facilities (central offices, POPs). Both are referenced in CDCP fire protection; 75 is the primary IT-equipment standard.

Aspirating smoke detection continuously draws air through a pipe network into a laser chamber, detecting sub-visible combustion products at the incipient stage — minutes or hours before a spot detector would alarm. NFPA 76 calls it out as best practice for high-value IT and telco spaces.

HFC clean agent that suppresses fire by heat absorption, leaves no residue, discharges in ~10 s, and is safe for occupied IT spaces. Trade-off: Global Warming Potential ~3220, so it is being phased down under the Kigali Amendment — long-term replacements (Novec 1230, IG-541) are preferred for new systems.

Fluoroketone clean agent with GWP close to 1 and atmospheric lifetime of days, making it the most environmentally favourable clean agent. Suppresses by heat absorption, is non-conductive, and is safe at design concentrations in occupied rooms. The recommended long-term replacement for HFCs like FM-200.

Inert-gas clean agents that suppress by oxygen depletion to ~12-15% (enough for humans to survive briefly, too low for fire to sustain). IG-541 = 52% N2 / 40% Ar / 8% CO2; IG-55 = 50% N2 / 50% Ar. They need large storage footprints (many cylinders) and longer discharge times (60-120 s) than chemical agents.

Multiple independent physical controls — perimeter fence, building shell, mantrap, room door, cabinet lock — so defeating one layer still leaves others. EN 50600-2-5 and ISO 27001 Annex A.7 codify this for data centres; each layer must be independently auditable and tested.

A two-door vestibule where the second door cannot open until the first has closed. Combined with weight scales, biometric readers, or anti-piggybacking sensors, it guarantees only one authenticated person enters the secure zone at a time, defeating tailgating behind an authorised colleague.

Biometric readers (fingerprint, iris, palm vein, facial) authenticate identity at the room or rack — harder to share or steal than badges. CCTV with analytics and EN 50131-classified intrusion detection record and alarm on perimeter or cabinet access, providing evidence and deterrence alongside the access-control decision.

MDA (Main Distribution Area) is the core cross-connect where external carriers meet the data centre backbone. HDA (Horizontal) feeds EDA (Equipment) cabinets via horizontal cabling. ZDA (Zone) is an optional consolidation point between HDA and EDA. The structured hierarchy keeps changes off the core and shortens MAC work.

Multimode uses LED/VCSEL light at 850-940 nm and cheaper optics but is distance-limited: OM3 ~70 m, OM4 ~100 m, OM5 (wideband, SWDM-optimised) extends SWDM4 reach beyond OM4. Single-mode uses laser sources, costs more per transceiver, but reaches kilometres — the choice for MDA-to-HDA and campus runs.

Multi-fibre Push-On terminates 12, 24, or 32 fibres in one ferrule, enabling parallel optics (40GBASE-SR4, 100GBASE-SR4 use 8 fibres per direction). MTP is US Conec's higher-performance brand of MPO. One MTP/MPO trunk replaces 12 LC jumpers, slashing cabling bulk in high-density spine-leaf fabrics.

SOP (Standard Operating Procedure) documents routine repeatable operations (start-up, daily checks). MOP (Method of Procedure) describes a specific change or maintenance task with step-by-step risk and rollback — required before non-routine work. EOP (Emergency Operating Procedure) defines response to fire, power loss, EPO activation, or flooding.

Cx 1: Factory witness test before shipment. Cx 2: Site inspection on delivery. Cx 3: Component verification — each unit tested individually post-install. Cx 4: Integrated Systems Test (IST) across the whole electrical/mechanical chain, including pull-the-plug fault simulation. Cx 5: Continuous or performance commissioning during operations.

BMS controls building HVAC, lighting, and life safety — coarse electrical visibility only. EPMS monitors electrical distribution down to branch-circuit and rack-PDU level with sub-cycle resolution. DCIM aggregates both plus IT asset and capacity data into a single pane of glass for capacity planning, asset tracking, and change management.

RTO (Recovery Time Objective) is the maximum acceptable duration a service can be unavailable after an incident — drives investment in failover speed and redundancy. RPO (Recovery Point Objective) is the maximum acceptable data loss measured in time since the last good replica — drives backup/replication frequency. A 1-hour RTO with a 15-minute RPO implies continuous replication and automated failover.

The Green Grid's PUE = Total Facility Energy / IT Equipment Energy. Perfect PUE = 1.0 (every watt reaches IT). Industry average sits near 1.55; hyperscale leaders report 1.10-1.20. PUE < 1.0 is impossible — values below 1.0 indicate a measurement boundary error. PUE is codified in ISO/IEC 30134-2.

WUE (Water Usage Effectiveness) = annual water consumed (L) / IT energy (kWh); Source-WUE adds upstream power-generation water. CUE (Carbon Usage Effectiveness) = annual CO2e (kg) / IT energy (kWh); REF (Renewable Energy Factor) = renewable energy / total energy. All are standardised in ISO/IEC 30134 (WUE part 9, CUE part 3, REF part 5).