DESIGN • CAPACITY PLANNING • DETAILED DESIGN
Data center infrastructure design
Concept design, permit and detailed design, capacity planning, CFD simulation and commissioning plans — electrical, cooling, rack, cabling, physical security and fire detection in one design team.
Design disciplines
Incoming power supply, UPS, PDU, single-line diagrams and the redundancy concept.
Cooling concept, load modeling and CFD airflow simulation.
Rack layout, power density, aisle containment and spare capacity.
Cable route design, cable trays, patch panels and a documented port schedule.
Zoning, access control and video surveillance design down to rack level.
Detection concept, aspirating smoke detection and rack-level fire detection design.
DESIGN PROCESS
Design documents and engineering methods
- Concept design and feasibility study
- Permit design and consultation with the authorities
- Detailed design and detail drawings
- As-built documentation
- Capacity planning and load modeling
- CFD airflow simulation
- Single-line electrical diagrams
- Redundancy concept (N+1, 2N)
- Tender documentation and technical specification
- Commissioning plan and commissioning protocol
FUNDAMENTALS
How is a data center design project structured?
A data center design project is not a single package of drawings but a sequence of phases that build on each other. It starts with requirements capture: what load has to be served today, what to prepare for in three to five years, what availability the business expects, and what the building allows. The concept design comes out of that, and it still thinks in options, each with a different capital and operating cost.
Once the concept is accepted, the permit design and consultation with the authorities follow, then the detailed design: the documentation you can actually build from. Tender documentation and the technical specification are assembled from it, so construction can be priced on a comparable basis, and finally the commissioning plan, which sets out how we will demonstrate that the built system does what the design promised.
It is worth being clear about where cost is decided. Not during construction: by then it only becomes visible what earlier decisions cost. The level of redundancy, the cooling concept, power density and expandability are fixed in the concept and detailed design phases, and these determine both the investment and the next ten years of operating cost.
The six disciplines — electrical, cooling, rack, structured cabling, physical security and fire detection — cannot be designed independently of each other. Rack power density drives the cooling requirement, the cooling solution takes up floor and ceiling space, cable routes compete for the same space, and fire detection has to be matched to the airflow. If six separate designers produce this, the clashes surface on site.
STANDARDS
What level should you design to? Tier levels and EN 50600
The first substantive decision in design is not technical but commercial: what availability the facility has to deliver. There are two widely used reference systems for this. The Uptime Institute Tier classification describes four levels, from a simple single-path supply (Tier I) to a fault-tolerant, fully duplicated architecture (Tier IV). The European EN 50600 series defines availability classes on a similar principle, and beyond that covers the full scope of designing, building and operating a facility.
Two key concepts are worth keeping apart: concurrent maintainability means any element can be maintained without interrupting the service; fault tolerance means an unexpected failure causes no outage either. The two are not the same, and the difference carries a significant difference in investment.
This translates into N+1 and 2N architectures. N+1 means one spare unit alongside the required capacity: good value, but it can leave common points. 2N means full, mutually independent duplication from the incoming supply to the rack. In practice a mixed solution is common: 2N on the electrical supply, N+1 on cooling.
PUE is decided in design, not in operations. The cooling concept, the aisle containment, the choice between air and liquid cooling and the power density together determine what PUE is achievable at all; from the measured value, operations can only fine-tune.
One important distinction: these frameworks give a design approach and a basis for comparison. Formal Tier certification or certification against a standard is a separate procedure, carried out by the certifying body concerned. If you need it, we align the design to it.
EXPERIENCE
Common mistakes in data center design
Capacity sized for day one. The design serves the initial load exactly, with no growth path. Two years later the first expansion runs into the incoming supply or the cooling, and the next step is no longer an expansion but a rebuild. Capacity planning has to include headroom and an expansion scenario as well.
Redundancy that exists on paper. The single-line diagram shows two independent paths, but in reality they run through one common distribution point, one cable route or one cooling circuit. Redundancy is only worth as much as its least duplicated point.
Cooling designed without CFD. Total cooling output is sufficient, yet the air still does not reach where it is needed. Part-load operation, the everyday reality for most data centers, is particularly unforgiving: systems sized for nominal load already produce hot spots at 40–60 percent part load.
Cabling designed last. Routes and trays come up only once the electrical and cooling solutions are already fixed. The result is congested floor space, obstructed airflow and a cable arrangement that no one dares disturb within a few years.
A missing commissioning plan. The system gets built, but there is no record of what tests it has to pass before handover. So no one demonstrates that load transfer, changeover and a cooling failure really behave as the design assumed; that surfaces at the first real incident.
Documentation that is never updated. The as-built documentation does not follow the changes made during construction, and later no one records the modifications. Every subsequent expansion starts with a fresh survey, and the design fee is paid again each time.
Planning a Server Room or Critical Infrastructure Project?
Briefly describe your needs, and one of our engineering colleagues will call you back soon.
or email us directly: info@digitechold.com
Environments we support
Enterprise data centers
Full infrastructure design from the incoming supply to rack level, with a redundancy concept.
Colocation data centers
Tenant separation, measurable capacity and expandability designed for multi-tenant environments.
Server rooms and edge sites
Building a server room within constrained conditions, with realistic spare capacity.
AI and high-performance computing (HPC) environments
Designing for high power density, liquid cooling and a reinforced power supply.
FAQ
Frequently asked questions about data center design
The concept design thinks in options: how much capacity, what redundancy, what cooling principle, and what capital and operating cost each carries. This is the decision-preparation phase. The permit design works the solution out to the extent the official procedure requires. The detailed design is documentation at the level you can actually build from: sized equipment, routes, junction details, wiring diagrams. The three build on each other, and producing a detailed design without an accepted concept is the most common source of re-pricing mid-project.
Typically the following: a technical description per discipline, sizing calculations and a capacity plan, single-line electrical diagrams, the cooling schematic and, where warranted, CFD simulation results, rack layout and route drawings, equipment and material schedules, tender documentation for pricing the construction, and a commissioning plan for handover. The project closes with as-built documentation recording what was actually built. That last one is what most places drop, and what they miss at every later expansion.
The length of the design phase depends primarily on the decision points, not on the drafting. A server room with a few racks can typically be taken from concept to detailed design in a few weeks. A data center project spanning several fire compartments and several disciplines takes considerably longer: evaluating concept options, consultation with the authorities and coordination between disciplines account for most of the time. Realistic planning starts with requirements capture, and we give a schedule from that before committing to any deadline.
Capacity planning pairs three numbers: today's actual load, foreseeable growth, and the maximum the infrastructure is able to serve. The starting point is power density per rack and how it is distributed, not aggregate kW, because hot spots are always local.
For headroom, we tend to recommend designing expandability rather than capacity built out unnecessarily: sizing the incoming supply, the routes and the space requirement so that the next step is an addition rather than a rebuild. Badly spent headroom costs more than none at all, and a system that cannot be expanded is the most expensive of the three.
The Tier classification expresses how far a facility tolerates maintenance and failure. From a design point of view, two concepts matter: concurrent maintainability means any element can be maintained without a service outage, and fault tolerance means an unexpected fault causes no outage either. The number of paths, the level of redundancy and how independent the branches are all follow from these. Important: formal Tier certification is a separate procedure with the certifying body. The design can be aligned to it, but the classification is not issued by the designer.
EN 50600 is a European series of standards covering data center facilities and infrastructure, and it uses availability classes along similar lines to the Tier logic. It covers a wider scope than the Tier classification: from the building structure through power supply, cooling, cabling and physical security to operational and energy efficiency considerations. In practice it gives the client, the designer and the contractor a good common language, because it translates the expected level into specific technical requirements instead of leaving “high availability” open to interpretation.
When the aggregate cooling output is not sufficient proof. Typically in three cases: at high or uneven power density, where hot spots are expected; when altering a live environment, where the existing airflow must not be disturbed; and when designing for part-load operation, because a significant share of systems work well at nominal load but no longer at 40–60 percent. For small, homogeneous server rooms it is generally not warranted, and there, carrying aisle containment and blanking panels through consistently is worth more.
A few well-defined factors drive the design fee: the size and capacity of the facility, the targeted availability level, how many disciplines the assignment covers, whether it is a greenfield investment or an upgrade carried out in a live environment, whether an official permit procedure is needed, and whether tender documentation, CFD simulation or a commissioning plan form part of the task. Designing in a live environment always means extra work, because the survey and the schedule have to fit around live operation. We provide a specific quote after requirements capture.
WHY US?
Why clients choose us
Six engineering disciplines in one design team. Electrical, cooling, rack, structured cabling, physical security and fire detection: the same team coordinates them, so clashes surface at the drawing board rather than on site.
Design and construction from one organization. The design is produced by the organization that can also build it. That makes the documentation not only correct but buildable, with realistic routes, procurable equipment and work sequences that can be scheduled.
A Tier and EN 50600-based design approach. We translate the availability expectation into specific technical requirements, instead of “high availability” meaning something different to everyone.
Survey and design in live, operating environments. A design produced in a working data center is a different task: the survey, the phasing and the schedule all have to fit around the service.
Support with consultation and permitting. We also take on assembling the permit documentation and answering the technical questions that come up during the procedure.
A commissioning plan as part of the design. We record at design time what tests the system has to pass before handover, so at handover it can be demonstrated that the built system does what the design promised.
Design and constructionfrom a single source.
Selected references
Data center and critical infrastructure design projects
Lenovo
Hungary
Citi
Hungary
Orlen
Hungary
Planning a data center or server room design project?
Tell us briefly where you are — requirements capture, concept, detailed design or an upgrade — and one of our engineers will come back to you within one business day.
- 5-minute project scoping
- Critical infrastructure expertise
- +36 30 835 2442