Q&A — Frequently Asked Questions About Data Centers

A Data Center (DC) is a centralized infrastructure designed to store, process, and distribute digital data. A DC includes servers, networking equipment, redundant power systems, cooling, physical security, and monitoring systems — all integrated to ensure continuous and secure operation for business data.

The Tier classification system defined by the Uptime Institute is based on the level of redundancy and continuous operational capability:

• Tier I: No redundancy, 99.671% uptime, suitable for small businesses.
• Tier II: Partial redundancy (N+1), 99.741% uptime.
• Tier III: Concurrent maintainability, 99.982% uptime — the most common today.
• Tier IV: Fault tolerant, 99.995% uptime, for mission-critical systems.

A Colocation Data Center is a model where you rent space and infrastructure from an external provider — the business only manages its own servers and data. An On-premise (self-built) Data Center is a model where the business invests in the entire infrastructure, providing absolute control but requiring significant capital investment and professional operational capabilities. The choice depends on the organization’s scale, budget, security requirements, and long-term strategy.

The design lifespan of a typical Data Center is 15–20 years. However, with the rapid pace of technological development, many DCs must upgrade or redesign their infrastructure after 8–10 years to meet increasing capacity and performance demands. Regular maintenance and a roadmapped upgrade plan are key factors in extending the operational lifecycle.

DC investment costs depend on several factors: the target Tier level, total IT capacity (kW), geographic location and regional technical infrastructure, selected cooling technology (air-cooled, liquid cooling), power system redundancy level, local labor and material costs, as well as international standard compliance requirements. On average, the cost of building a DC ranges from 8–15 million USD/MW depending on scale and tier.

Key international standards include: Uptime Institute Tier Standard (redundancy classification), TIA-942 (telecommunications infrastructure for DCs), EN 50600 (European standard for DCs), ASHRAE TC 9.9 (temperature and humidity guidelines), IEC 60364 (low-voltage electrical installations), NFPA 75/76 (fire protection), and ISO/IEC 27001 (information security).

Power and cooling systems are two factors that directly determine the Tier level. Tier III requires independent power and cooling paths (multiple active paths), allowing maintenance without service interruption. Tier IV requires each redundant path to be fully fault-tolerant, ensuring no single point of failure (SPOF). A design that fails to meet redundancy requirements in either of these systems will directly impact the Tier certification.

N+1 redundancy means the system has one extra backup unit beyond what is required for operation (N). 2N redundancy means the entire system is fully duplicated. The goal is to ensure that when a piece of equipment fails or requires maintenance, the remaining system continues to operate normally — eliminating downtime risks and protecting service continuity.

PUE (Power Usage Effectiveness) is a metric used to measure the energy efficiency of a DC, calculated by dividing the total energy consumed by the entire DC by the energy consumed by the IT equipment. An ideal PUE is 1.0 (all power is used for IT). In reality: a PUE ≤ 1.2 is considered excellent, 1.2–1.5 is good, 1.5–2.0 is average, and anything above 2.0 requires significant improvement.

Key measures include: designing adequate Tier-level redundancy, developing and strictly adhering to SOPs, performing regular preventive maintenance (PPM), periodically testing backup systems (UPS, generators, ATS), deploying real-time monitoring systems (DCIM/BMS), providing regular operational team training, and establishing a clear Incident Response Plan (IRP).

Maintenance frequency depends on each system: UPS and batteries — monthly checks, comprehensive maintenance every 6 months; diesel generators — monthly load tests, routine maintenance based on operational hours; cooling systems (chillers, CRACs) — maintenance every 3–6 months; fire protection systems — quarterly checks; cables and connections — annual checks. The maintenance schedule must be documented in a PPM plan and fully archived.

SOP (Standard Operating Procedure) is a set of documented standard operating procedures that specify step-by-step instructions for every activity in the DC — from daily operations to troubleshooting and maintenance. SOPs help minimize human error, ensure operational consistency, assist in training new personnel, and serve as a basis for audits and Tier certification.

A DRP (Disaster Recovery Plan) should include: Business Impact Analysis (BIA), determining RTO (Recovery Time Objective) and RPO (Recovery Point Objective), developing response scenarios for various disasters (power outages, fires, cyberattacks, natural disasters), clear assignment of responsibilities, procedures for activating the backup DC, and regular drill schedules to test the plan’s effectiveness.

T&C (Testing & Commissioning) is the process of testing and accepting all technical systems of a DC before commercial operation. This phase verifies that all systems — power, cooling, UPS, generators, fire protection, BMS — operate according to design, integrate synchronously, and meet Tier-level requirements. Skipping T&C means accepting uncontrollable operational risks.

The T&C process typically consists of 5 phases: (1) Factory Acceptance Test (FAT) — testing equipment at the manufacturing facility; (2) Site Acceptance Test (SAT) — testing equipment on-site after installation; (3) Integrated System Test (IST) — comprehensive system integration testing; (4) Load Testing — testing under actual load conditions; (5) Final Commissioning & Handover — final acceptance and handover of operational documentation.

T&C is a collaborative responsibility among multiple parties: The contractor is responsible for performing the tests and providing evidence of the results; the supervising consultant (or an independent T&C firm) is responsible for verifying and witnessing the test results; the investor has the right to request re-testing if the results are unsatisfactory. Hiring an independent T&C firm is a best practice to ensure objectivity.

The T&C period for a medium-sized Tier III DC (1–5MW IT load) usually takes from 8–16 weeks, depending on system scale, complexity, and the number of test scenarios required. The IST and load test phases usually take the most time. Detailed T&C planning from an early stage and close coordination among contractors are crucial factors for the timeline.

A Green Data Center is a data center model designed and operated to optimize energy efficiency, minimize carbon footprint, and reduce environmental impact. Characteristics of a Green DC include: a low PUE, use of renewable energy, highly efficient cooling systems, waste heat reuse, smart water management, and achieving international environmental certifications such as LEED and ISO 50001.

LEED (Leadership in Energy and Environmental Design) is the USGBC’s green building certification system, applied to DCs in categories: energy efficiency, water management, sustainable building materials, indoor environmental quality, and design innovation. A DC can achieve LEED certification at 4 levels: Certified, Silver, Gold, and Platinum — corresponding to the points earned in the rating system.

Common solutions include: deploying hot aisle/cold aisle containment to optimize airflow; using high-efficiency cooling systems (economizers, free cooling, liquid cooling); raising operating temperatures according to ASHRAE guidelines (18–27°C); applying AI/ML to optimize cooling system operations; utilizing new generation UPS with >96% efficiency; and deploying DCIM systems to monitor and optimize energy in real-time.

Renewable energy can be integrated into DCs through various methods: installing solar PV systems on the roof or surrounding areas; signing direct Power Purchase Agreements (PPAs) with renewable providers; purchasing Renewable Energy Certificates (REC/I-REC); and participating in smart grids. The current trend is for large DCs to commit to 100% renewable energy following a specific roadmap.

DC physical security is organized according to a defense-in-depth model: Layer 1 — Perimeter (fences, CCTV cameras, vehicle control); Layer 2 — Building (access control via keycards/biometrics); Layer 3 — Data Hall (strict access control, mantrap/airlock); Layer 4 — Cabinet/Rack (rack locks, device-level control). Each layer requires independent policies, equipment, and control procedures.

A DC fire protection system typically includes: an early warning system (VESDA — Very Early Smoke Detection Apparatus), a multi-zone fire alarm system, and a clean agent fire suppression system (FM200, Novec 1230, or CO2) instead of water to prevent equipment damage. The gas discharge process must be integrated with commands to shut down the HVAC system and trigger personnel evacuation warnings before activation.

In Vietnam, DCs must comply with: QCVN 06:2022/BXD (national technical regulation on fire safety of buildings and constructions), TCVN for electrical and mechanical systems, Ministry of Information and Communications regulations on DC infrastructure, and Decree 13/2023/ND-CP on personal data protection. Additionally, DCs targeting the international market often comply with ISO 27001, SOC 2, and PCI-DSS.

The TCO of a DC includes: Initial Capital Expenditure (CapEx) — land, construction, IT equipment, and infrastructure; Annual Operational Expenditure (OpEx) — power (accounting for 40–60% of OpEx), operational personnel, maintenance, transmission line leasing, insurance; Upgrade and replacement costs over the lifecycle; Compliance and certification costs. Analyzing the full lifecycle TCO (15–20 years) is a crucial basis for comparing investment options.

You should build an On-premise DC when: the organization has a high capacity demand (>1MW), requires comprehensive control over security and compliance, possesses adequate technical operation capabilities, and has a long-term vision (>10 years). You should choose Colocation when: the capacity demand is small or fluctuating, quick deployment is needed, you want to optimize initial capital investment, or you lack a professional technical operation team. A hybrid model (combining both) is becoming increasingly popular.

Strategies to optimize long-term operational costs include: improving PUE through cooling system upgrades and airflow optimization; transitioning to renewable energy to reduce power costs; deploying DCIM to detect energy waste and optimize infrastructure usage; performing preventive maintenance to reduce emergency repair costs; and applying virtualization and consolidation to optimize IT equipment density.