EXIN CDCS: Data Centre Design, Standards, and Operations
Domain scope
EXIN CDCS (Certified Data Centre Specialist) tests whether you can apply data-centre standards, redundancy logic, and operational controls to realistic facility decisions. Questions typically ask you to classify a design, interpret a power or cooling topology, or choose the correct operational artefact.
Core rules and concepts
- Standards map. ANSI/TIA-942-B uses Tier I–IV (basic, redundant, concurrently maintainable, fault tolerant). EN 50600 uses availability classes 1–4; ISO/IEC 22237 is the international series. Do not mix Tier numbers with EN classes.
- Redundancy notation. N is the capacity needed; N+1 adds one unit; 2N duplicates the full path; 2(N+1) duplicates each N+1 path. Redundancy is about independent paths, not just spare modules.
- Power chain. Utility → MV switchgear → transformer → LV switchgear → ATS/STS → UPS → PDU → rack PDU. A true 2N design has independent A and B paths to dual-corded IT equipment.
- Cooling and ASHRAE. ASHRAE TC 9.9 recommends an inlet envelope (commonly 18–27°C, with humidity limits); containment separates hot and cold air; economizers/free cooling reduce compressor use. PUE = total facility energy ÷ IT energy.
- Fire and security. VESDA gives very-early smoke detection; pre-action or clean-agent suppression protects IT spaces; mantraps and access control enforce security zones.
- Operations. MOP, SOP and EOP are method, standard and emergency operating procedures; change management and DCIM records support capacity and risk control.
Worked example
IT load is 800 kW. Each UPS module is rated 400 kW and site policy caps module loading at 80%. For N+1, usable capacity with one module failed must still cover 800 kW. At 80%, each healthy module delivers 320 kW. Two modules give 640 kW (insufficient); three give 960 kW (sufficient). Therefore the design needs 3+1 = 4 modules, not 2+1. The trap is checking nameplate N+1 without applying the loading cap.
Traps that cost marks
- Calling N+1 “fault tolerant”; fault tolerance requires 2N or equivalent independent paths.
- Treating Tier III as “no downtime”; it is concurrently maintainable, not fault tolerant.
- Reporting PUE as a percentage or saying higher PUE is better.
- Ignoring concurrent maintainability when only one UPS or switchgear train exists.
- Assuming clean-agent discharge is automatic on a single smoke detector; usually two alarms and a pre-discharge delay are required.
- Confusing an ATS (source transfer) with an STS (load transfer).