Datacenter
Data center white space and gray space layout field guide
What white space and gray space are, why the data hall and the support plant get split, how the rooms lay out, and what actually caps the capacity before the floor runs out.
Direct answer
White space is the data hall where IT racks and servers sit, the raised floor or slab area built for the computing load. Gray space is the back-of-house area for the support gear, the UPS, switchgear, generators, and cooling plant that keeps the white space running. The project program and the TIA-942 spaces control the split.
Key takeaways
- White space is the conditioned data hall holding IT racks and servers; gray space is the back-of-house mechanical and electrical plant that powers and cools it.
- Capacity is the smallest of three limits: floor area, power, and cooling; on a modern floor space is almost never the constraint that runs out first.
- TIA-942 recommends a cold aisle on the order of 4 ft (about 1.2 m), a two-tile width on a raised floor; egress code overrides the airflow plan where they disagree.
- Run the redundant A and B electrical and mechanical rooms physically separate end to end so one fire or flood cannot take both.
- TIA-942 governs facility zoning and recommends locating the entrance room outside the computer room; cabling labeling follows TIA-606.
White space and gray space, the two halves of the building
White space is the part of a data center where the IT gear lives. It is the data hall, the raised floor or slab room full of racks, servers, storage, and network gear, laid out in rows with the cooling and power that feed them. Gray space is everything behind it that keeps it alive: the support area for the mechanical and electrical plant, the UPS, the switchgear, the batteries, the generators, the chillers, and the pumps. One zone runs the computing. The other zone runs the building.
The two names come from how the rooms read. The data hall is bright, finished, and clean, the white space. The support areas are raw mechanical and electrical rooms, the gray space. The line between them is not decoration. It sets who has access, who maintains what, how the air and power move, and where the security and fire boundaries fall.
This guide covers the zoning, the layout, and the rooms. It leans on two companions and does not repeat them. The rack readiness and floor load guide covers what happens inside the white space at the cabinet, the placement, the floor rating, the bonding, and the readiness gate. The power distribution chain guide covers the electrical side of the gray space, every stage from the utility to the rack. This one is the building map that holds both.
Why separate the white space from the gray space?
The split exists because the IT load and the gear that supports it share nothing except dependence. The white space wants tight temperature and humidity control, restricted access, and a clean, quiet environment. The gray space holds loud, hot, oily, high-energy machinery that needs service access, ventilation, and room to pull a UPS module or a breaker. Put them in the same room and each one fights the other.
Access is the first reason. A mechanical contractor servicing a chiller has no business standing in a customer's data hall, and a customer racking a server has no business near live medium-voltage switchgear. Separating the zones lets each crew work its own space without escorting it through the other. In a colocation building the whole security model depends on this: the customer reaches the white space without ever touching the gray space, and the facilities team reaches the gray space without ever touching a customer's racks.
Maintenance and risk are the second reason. The gray-space gear gets worked on live, replaced, and load-tested on a schedule. Keeping that work out of the white space means a battery replacement or a generator test does not happen over the top of running servers. The fire and the failure modes differ too, so the detection and suppression are designed per zone, which is far easier when the zones are walled apart.
What is white space in a data center?
White space is the conditioned IT area, the data hall where the racks and rows of equipment sit and run. It is the floor the whole building is built to serve, and it is measured and sold by its area and, increasingly, by the power and cooling it can deliver to that area. Everything else in the facility exists to keep this room inside its temperature, power, and uptime envelope.
The white space holds the cabinets in rows, arranged so the equipment intakes face a cold aisle and the exhausts face a hot aisle. It sits on either a raised access floor or a slab, with power arriving by busway or whip and cooling arriving by perforated tile, in-row unit, or overhead. The rack readiness guide covers the cabinet-level detail, the placement on a grid coordinate, the floor load, the bonding, and the containment that make a rack ready for gear.
What defines the white space is the control of its environment. The air is held in a tight band, the floor is sealed against leakage, and access is limited to the people and the work that belong there. A clean, sealed, coordinate-mapped data hall is white space. A room full of racks with no environmental control and no access boundary is just a server closet.
What is gray space in a data center?
Gray space is the back-of-house support area that holds the mechanical and electrical plant. It is where the gear that powers and cools the white space lives: the UPS and its batteries, the switchgear and transformers, the floor distribution, the standby generators, the chillers, the pumps, the air handlers, and the controls. None of it computes anything. All of it keeps the white space running.
The name marks the difference in finish and function. Gray space is mechanical and electrical room space, designed for the equipment and the people who service it, not for IT gear. It needs heavy floor loading for transformers and switchgear, ventilation for the heat the gear throws, working clearances in front of electrical equipment, and routes wide enough to roll a UPS module or a breaker truck in and out.
The power distribution chain guide walks the electrical side of the gray space stage by stage, from the utility service through the switchgear, the UPS, and the floor distribution. The point here is the zoning: the gray space is a deliberate, separate part of the building, sized to the white space it carries, and its footprint grows with the density of the load.
The white-to-gray ratio and where it is going
The ratio of white space to gray space is one of the numbers that describes a data center, and it has been moving against the white space for years. Older, low-density halls supported a lot of IT floor with relatively little support area. As the power per rack climbed, the gear needed to feed and cool that rack grew, so each square foot of white space now demands more gray space behind it.
There is no single correct ratio, because it depends on the density, the redundancy, and the cooling approach. A lean, low-redundancy, air-cooled hall carries less gray space per unit of white space than a 2N, high-density, liquid-cooled build. A higher Tier or redundancy target multiplies the electrical and mechanical gear, which multiplies the gray space. Treat any published ratio as a rough planning figure, not a design rule, and size the gray space from the actual load and redundancy.
Liquid cooling pulls the trend in both directions. It lets a given white-space footprint hold far more compute, which raises the load per square foot of data hall. But the coolant distribution units, the heat rejection, and the piping all need their own room and routes, so the support footprint grows too. The lesson for layout keeps repeating: the gray space is sized by the load, and the load per rack is still climbing.
Raised access floor or slab on grade?
A raised access floor is a grid of removable panels on pedestals, set above the structural slab, with the gap underneath used as a plenum for cooling air and, historically, for power and cabling too. Slab on grade puts the racks on the structural floor and runs the power, cooling, and cabling overhead. Both are in service today, and the trade has been shifting toward slab with overhead distribution on new high-density builds.
The raised floor earned its place by using the underfloor void to push cold air up through perforated tiles into the cold aisle. That works well at moderate density. The trouble is that the same void fills with cabling and power that block the airflow it exists to deliver, every cable penetration is a leak that has to be sealed, and the pedestals cap the point load the floor can take. A loaded high-density rack or a liquid-cooled cabinet is a structural problem on a raised floor that a slab handles without thinking.
Slab with overhead has gained ground because high-density cooling does not need the plenum. In-row units, rear-door heat exchangers, and direct-to-chip liquid cooling deliver cooling at or in the rack, so the underfloor void buys nothing. The slab takes the weight directly, the anchoring is simpler, and there is no plenum to leak. The rack readiness and floor load guide covers the load-rating side of the choice. The layout point is that the floor decision drives where the power, cooling, and cabling run, so it is made early and it is hard to reverse.
What is hot aisle and cold aisle layout?
Hot aisle and cold aisle is the row arrangement where cabinets are turned so equipment fronts face fronts across a cold aisle and backs face backs across a hot aisle. Cold supply air is delivered to the cold aisle where the intakes draw it, the gear pulls it through and dumps hot exhaust into the hot aisle, and the cooling system collects that exhaust and rejects the heat. The whole point is to keep the cold supply and the hot exhaust from mixing.
Turn one cabinet the wrong way and it breaks the pattern: it pulls its neighbor's hot exhaust into its own intake and runs hot. The rack readiness guide covers the cabinet-level discipline, the blanking panels, the seals, and the containment that hold the separation at each rack. The layout-level job is to set the row orientation across the whole hall so every cold aisle faces cold and every hot aisle faces hot, with no flipped cabinets and no mixed rows.
Containment takes the idea further by physically enclosing one aisle. Cold-aisle containment caps and doors the cold aisle so the conditioned air stays pinned to the intakes. Hot-aisle containment encloses the hot aisle and ducts the exhaust back to the units. Either way the separation gets tighter, the supply setpoint can rise, and the cooling energy drops. The choice between them is a design call that the layout, the ceiling, and the floor approach all have to agree on.
Aisle widths and service clearances
Aisle widths come from three demands at once: the airflow, the service access, and the code egress. The cold aisle has to be wide enough to deliver the air and to work in front of the racks. TIA-942 gives a recommended cold aisle on the order of 4 ft (about 1.2 m), which on a raised floor lands on a two-tile width. The hot aisle is usually set by the rear door swing and the room to service a loaded cabinet from behind.
Service clearance is real and it is loaded, not nominal. A deep server plus its cable management plus the door swing has to leave the aisle passable with the doors open. In the gray space, the electrical clearances are stricter and enforceable: the working space in front of switchgear, panels, and PDUs is a code requirement, not a preference, and it has to stay clear. Confirm the working-clearance dimensions against the adopted electrical code for the voltage and condition.
Egress runs underneath all of it. The aisles double as the means of escape from the room, so the egress widths and the travel distances to an exit follow the building and life safety codes regardless of how tight the airflow plan would like to pack the rows. Where the airflow plan and the egress code disagree, the egress wins. Hedge every width to the project drawings and the adopted code, because they vary by jurisdiction and by the room's classification.
Rows, the floor grid, and the coordinate system
Every cabinet position is placed on a grid before any rack lands, and the grid is what keeps the whole hall coordinated. On a raised floor the grid keys to the access tiles, commonly 600 mm (about 24 in), lettered one way and numbered the other so a position reads like a square on a board. On slab the grid is painted or surveyed onto the floor to the same effect. The rack, its power circuits, its cooling, and its cabling all reference that one coordinate.
The row plan is the drawing that places each cabinet on the grid, fixes the aisle pitch, and sets which way every rack faces. It is also where the containment, the overhead tray, the busway, and any tile cuts have to agree. The rack readiness guide goes deep on holding the layout to the coordinate instead of a tape off the wall, because walls are rarely square and the drift over a long row pushes the last cabinet off its taps and cuts.
The naming carries past the white space. The labeling and administration of the spaces and the cabling follows TIA-606, so the grid coordinate, the rack ID, the power circuit, and the cable labels all tie to one scheme that the records and the steel agree on. Set the convention before the first label prints. A hall where the coordinate system is consistent from the floor tile to the patch panel is a hall you can manage. One where every trade invented its own is a permanent scavenger hunt.
Getting power and cooling from the gray space to the racks
The whole reason for the zoning is that the power and the cooling are made in the gray space and consumed in the white space, so the layout has to carry both across the boundary cleanly. On the power side, the protected output from the UPS and the floor distribution reaches the rows by overhead busway or by whips from a floor PDU or a remote power panel. Busway has taken over the dense floor because it runs overhead, leaves the cooling path clear, and lets a rack feed move or grow with a plug-in tap instead of a new conduit run.
On the cooling side, the heat made in the white space has to get back to the gray space or to the outdoors to be rejected. Perimeter units, the CRAC and CRAH, sit at the edge of or just outside the hall and move air. In-row units sit between the cabinets and cool close to the load. Rear-door heat exchangers and direct-to-chip systems pull the heat into a water loop through a coolant distribution unit. Every one of those needs a path back to the chilled water or the heat rejection in the gray space.
The power distribution chain guide details the electrical stages, and the rack readiness guide details how the feeds and the cooling land at the cabinet. The layout point is the routing. The busway, the piping, the tray, and the air all share the overhead and the room boundary, and they have to be coordinated so they do not fight each other or block the airflow. A hall that ran its cabling and power under a raised floor and choked the plenum is the classic version of getting this wrong.
The electrical rooms in the gray space
The electrical side of the gray space is a series of rooms that step the power down and protect it on the way to the floor. The lineup typically runs from the medium-voltage service and switchgear, through the transformers, to the low-voltage switchgear, the UPS and its batteries, and the floor distribution that feeds the white space. Each gets its own room or bay, sized for the gear and for the clearances to work on it safely.
The battery room is its own concern. A UPS holds its ride-through energy in batteries or flywheels, and a battery room carries ventilation, spill, and sometimes hydrogen requirements that drive its own design. The switchgear rooms need the working clearances and the egress that high-energy electrical work demands, plus the arc-flash boundaries that come with it. These are not rooms you walk through casually, which is part of why they sit in the gray space behind their own access boundary.
The power distribution chain guide is the place for the stage-by-stage electrical detail. For the layout, the thing to hold is that these rooms are sized and placed to feed the white space on the shortest sensible path, with the redundant A and B electrical rooms kept physically separate so a fire or a flood in one does not take both. Two electrical rooms that share a wall, a route, or a roof leak path are less independent than the one-line suggests.
The mechanical plant and where it sits
The mechanical plant is the cooling side of the gray space, and a good part of it lives outside the building entirely. Inside, the gray space holds the pumps, the air handlers, and the controls. Outside or on the roof sit the chillers, the cooling towers, the dry coolers, or the condensers that reject the heat to the air. The chilled water or the refrigerant loops tie the outdoor heat rejection to the indoor units that condition the white space.
The plant is sized to the IT load plus the losses, and it carries its own redundancy. An N+1 or 2N chilled-water plant means a chiller or a pump can drop or be serviced with the cooling still holding, the same redundancy logic the electrical side uses. The cooling cannot be allowed to stop, because a high-density hall overheats in minutes, not hours, when the air or the water stops moving.
The layout has to give the plant its routes and its room. The piping risers, the air paths, and the heat rejection all need space and a path that does not cross the electrical rooms or the white space where it should not. As liquid cooling spreads, the piping into the white space grows, the coolant distribution units need floor space near the rows, and the heat rejection outside grows with the load. The mechanical footprint is climbing alongside the electrical one.
What is a meet-me room?
A meet-me room (MMR) is the space where the carriers and networks in a data center physically interconnect. It is the cross-connect point where a customer in the white space reaches the outside world, and where carriers exchange traffic with each other inside the building instead of routing it out and back. In a colocation or carrier-hotel facility, the MMR is the connectivity hub the whole tenant base depends on.
It pairs with the entrance facility, the entrance room where the outside cables actually enter the building. Carrier fiber comes in through the entrance facility, often from diverse routes and diverse entry points so one cut does not isolate the building, and lands in or near the MMR. TIA-942 recommends locating the entrance room outside the computer room for security, which is the zoning logic at work: the place outside cables enter is its own controlled space, separate from the white space and the gray space.
The MMR is treated as a high-security zone in its own right, often in fire-rated walls with its own access control, because it is the single point where cutting or tapping connectivity would hit every tenant at once. For resilience, larger buildings run two MMRs in separate locations so the interconnection itself has an A and a B. Where the MMR and the entrance facility sit relative to the white space is one of the first decisions in the building layout, because every customer's connectivity routes through it.
Staging, build, and storage rooms
A working data center needs room to receive, build, and stage gear before it ever reaches the white space, and that space is easy to leave off the program until the first truck arrives. Receiving and a loading dock take the deliveries. A staging or build room is where racks get assembled, gear gets mounted, and equipment gets configured and burned in before it rolls onto the floor. Storage holds spares, blanking panels, tiles, and the parts a running facility consumes.
The reason to keep this work out of the white space is the reason the zones exist at all. Unpacking cardboard, drilling, and burning in new gear is dirty, particle-heavy work that does not belong in a clean, contained data hall. Cardboard and packing material are also a fire load you keep off the floor deliberately. Staging the rack in a build room and rolling in a finished, tested cabinet protects the environment and the schedule both.
Burn-in matters because new gear tends to fail early or not at all. Running equipment under load in a staging area for a period before it goes live catches the infant-mortality failures off the production floor, where a failure is cheap, instead of after it is racked and serving load. A facility with no staging area ends up doing all of this in the white space or the aisles, which is exactly where you do not want it.
How is a data center zoned for security?
A data center is secured in layers, like rings, so that getting deeper into the building takes clearing one more boundary each time. The outer ring is the site: the fence, the gates, and the vehicle control. Inside that is the building shell and the lobby, where people are identified and badged. Past the lobby is the gray space and the general interior. Then the white space. Then, in a colocation building, the individual customer cage or suite. Each ring is a separate access decision.
The mantrap is the device that enforces the boundary at the sensitive rings. It is a pair of interlocked doors where the second will not open until the first has closed and the person has cleared a second factor, a badge plus a biometric or a PIN, so one credential alone does not pass and a second person cannot tailgate through. Access to the white space and the cages typically passes through a mantrap, often with anti-tailgating sensors that lock down if two people try to enter on one clearance.
The zoning maps onto the white and gray split directly. A colocation customer is granted the lobby and the path to their own cage in the white space and nothing else. They never reach the gray space or another tenant's cage. The facilities team reaches the gray space and the plant and does not need the customer cages. The access control system is built around those rings, and the boundary between them is the same boundary the rest of the building design is organized on.
Cages and colocation suites in the white space
In a colocation building, the white space is divided among customers, and the unit of division is the cage or the suite. A cage is a fenced, lockable enclosure around a customer's rows inside the shared data hall, giving them a private, access-controlled area on the common floor. A suite is a walled private room for a larger customer who wants their own four walls. Either way, the customer's gear sits inside their boundary, on the shared floor, fed by the shared power and cooling.
The cage is a security and a billing boundary at once. It limits physical access to the tenant's own staff and their escorted visitors, separating one customer's gear from the next on a floor full of competitors. The layout has to lay the cages out so each one gets its access aisle, its power feeds, and its cooling without crossing another tenant's space, which is a real constraint on how the rows and aisles are planned.
Cages also change the airflow and the containment problem. A cage mesh or a suite wall can block or redirect the air the open floor was designed to move, so the containment and the tile or unit placement have to account for the boundary. A cage dropped onto a floor that was planned as open white space, without rechecking the airflow, is a common way to strand cooling on one side of the mesh.
Fire and life safety zoning
Fire detection and suppression are designed per zone, because the white space and the gray space burn and fail differently. The white space is a high-value room full of always-running equipment where a discharge is expensive and a false trip is costly, so it commonly uses very early smoke detection and a suppression approach chosen to protect electronics. The gray space holds fuel, oil, batteries, and high-energy gear with their own hazards and their own suppression needs.
Very early smoke detection, the aspirating systems that sample the air continuously, is common in the white space because catching a hot component before it flames lets staff act before suppression ever has to discharge. Suppression itself ranges from pre-action sprinkler systems, kept dry until a detector confirms a fire so a single broken head does not soak the racks, to clean-agent gas systems that put a fire out without water in the room. Which one a room uses is a design and code decision tied to the room and its contents.
Life safety runs alongside the fire protection and does not bend to the airflow plan. The egress paths, the exit travel distances, the emergency lighting, and the fire-rated separations between zones follow the building and fire codes, the NFPA suite among them, for the occupancy and the hazard. The fire-rated walls that separate the white space, the gray space, the battery rooms, and the meet-me room are part of the zoning, not just partitions. Confirm the detection, the suppression, and the rated separations against the adopted codes and the AHJ, because they vary by jurisdiction and by the room's classification.
What actually limits a data center's capacity?
Capacity is limited by whichever of power, cooling, or space runs out first, and on a modern floor it is almost never the space. A data hall is planned on three capacities at once: the floor area, the power it can deliver, and the cooling it can reject. The usable capacity is the smallest of the three, because a rack position with no power to feed it or no cooling to keep it is a position you cannot fill.
Stranded capacity is the name for the gap. It is installed capacity that cannot be used because another resource ran out next to it. Floor space with no power left to feed it is stranded space. Power that cannot be used because there is no cooling where it is needed is stranded power. A floor that fills to its power limit at half its physical racks has stranded the other half of its space. The three have to be planned together, because optimizing one in isolation strands the others.
The shift that broke the old habit is density. When racks pulled a few kilowatts, the floor area was usually the limit and you could plan on space. As the power per rack climbed into the tens and hundreds of kilowatts, power and cooling became the binding constraints, and a hall now runs out of one of them long before it runs out of floor. Plan the capacity on the power and the cooling, size the gray space to deliver them, and treat the floor area as the constraint that usually is not the one that bites.
How high density reshapes the layout
High-density and AI compute have changed the layout, not just the gear. A traditional hall spread a moderate load evenly across a lot of racks. An AI build concentrates enormous power and heat into a few rows, which changes where the power, the cooling, and the weight have to go. The layout follows the concentration: the high-density rows get the heavy busway, the liquid cooling, and the structural floor that can take the weight.
The cooling is the first thing to change. Air alone runs out somewhere in the tens of kilowatts per rack, so the dense rows move to liquid, which brings coolant distribution units onto the floor, piping into the white space, and heat rejection capacity into the plant. The rack readiness guide covers the cabinet side of liquid cooling. The layout side is that a liquid row needs floor space for the CDU, routes for the piping, and leak detection designed into the room.
The power and the structure change with it. The dense rows pull higher-current, higher-voltage feeds, which pushes the busway ratings up and the gray-space electrical capacity up to match. The concentrated weight of packed, liquid-cooled cabinets pushes the floor decision toward slab. A building laid out for even, moderate density does not become an AI hall by swapping the racks. The power, the cooling, and the structure all resize, and the gray space grows to carry it.
Commissioning and turning over the space
The white space and the gray space get commissioned together, because neither one proves out alone. The gray-space gear is tested stage by stage and then as an integrated system, the power chain pulled under load and the cooling proven to hold while a failure is forced. The white space is proven for its environment, its containment, and its readiness to take gear. The handoff is the point where the construction and commissioning side proves the space does what the design promised before the IT load arrives.
A load bank is how the space gets proven without real servers. Banks of resistive load are set in the white space to simulate the IT heat, the power chain carries it, and the cooling rejects it, so the whole facility runs at design load before a single customer server is at risk. This is where an undersized stage, an A and B path that secretly share a source, or a cooling system that cannot hold setpoint at full load get found, while the only thing on the floor is a load bank.
The rack readiness guide covers the cabinet-level gate, the per-rack checklist that has to clear before IT loads a cabinet. The power distribution chain guide covers the integrated electrical test. The facility-level point is that the zones turn over as a coordinated whole. A white space that is ready but sits on a gray space that never passed its integrated test is not ready, and the order of testing matters: prove the support before you trust it with the load.
How the whole building is organized
Step back from the rooms and a data center building sorts into three broad zones. The administrative and people space, the offices, the lobby, the security operations center, and the staff areas, sits at the front where people enter. The gray space, the electrical and mechanical plant, wraps the support gear around or alongside the data halls. The white space, the data halls themselves, sits at the protected core, reached only through the security rings and fed only by the gray space.
The arrangement is driven by flow and by separation. People flow from the front, through the lobby and the security boundary, toward the work they are cleared for. Power and cooling flow from the gray space into the white space on the shortest sensible path. The outside utilities, the electrical service and the carrier fiber, enter at controlled points, the service yard and the entrance facility, on the edge. The deliveries flow through receiving and staging before anything reaches the floor.
Good layout keeps those flows from crossing where they should not. The customer path never crosses the gray space. The mechanical service path never crosses the white space. The redundant A and B systems stay physically apart. A building that gets the zoning right reads cleanly: you can stand at any door and know which zone you are in and what is on the other side. One that grew without the discipline becomes the building where the only person who knows where anything is is the one about to retire.
What to document for the zoning
The record that matters for the layout names every zone, what it holds, and the concern that governs it, so the next person can read the building without walking every room. Capture the white space and its environmental and load envelope, the gray-space rooms and their gear and clearances, the meet-me room and entrance facility, the security rings, and the fire and life safety zoning. A building whose zoning lived only in the head of the original designer is a building you re-survey every time something changes.
| Zone | What it holds | Key concern |
|---|---|---|
| White space / data hall | IT racks, rows, containment, raised floor or slab | Environment, power and cooling per rack, access |
| Gray space electrical | UPS, batteries, switchgear, transformers, distribution | Working clearances, redundancy, A and B separation |
| Gray space mechanical | Pumps, air handlers, CDUs, controls, heat rejection | Cooling redundancy, piping routes, plant capacity |
| Meet-me room / entrance facility | Carrier interconnect, incoming fiber | Diverse entry, high security, redundant MMR |
| Staging / receiving | Dock, build room, burn-in, storage | Keep dirty work and fire load off the floor |
| Security zones | Lobby, mantraps, cages, access rings | Layered access, tailgating, tenant separation |
| Life safety | Egress, suppression, rated separations | Code egress, per-zone suppression, AHJ signoff |
Common mistakes
- Sizing the gray space for today's density and running out of electrical and mechanical room as the load per rack climbs.
- Planning the capacity on floor area and stranding it on power or cooling that ran out first.
- Laying out the white space with flipped cabinets or mixed rows that break the hot-aisle and cold-aisle separation.
- Choosing a raised floor where slab and overhead would have carried the weight and the high-density cooling better.
- Building no staging or build room, so unpacking, assembly, and burn-in end up in the white space or the aisles.
- Running the redundant A and B electrical or mechanical rooms close enough that one fire or flood takes both.
- Treating the meet-me room and entrance facility as an afterthought with a single entry route and weak access control.
- Weak security zoning that lets a colocation customer or a contractor reach a zone they were never cleared for.
Field checklist
Want this checklist to run itself on every job — with photo proof and a signed record crews can hand the customer? That's FieldOS.
Standards and references
TIA-942 is the data center infrastructure standard, and it is the document the facility zoning traces back to. It defines the functional spaces, the computer room (the white space), the entrance room where outside cabling lands, the main and horizontal distribution areas, and the support spaces, and it recommends separating them, including locating the entrance room outside the computer room for security. It also frames the white-space layout, the hot-aisle and cold-aisle arrangement, and the cabling and labeling, which follows TIA-606.
The Uptime Institute's Tier classification is the common language for how much of the facility can fail or be maintained with the load still up, which drives how much redundant electrical and mechanical gear the gray space has to hold. ASHRAE Technical Committee 9.9 publishes the thermal guidelines the cooling and the white-space environment are designed to, the recommended server inlet temperature and humidity envelope that the containment and the aisle layout exist to hold.
On the building side, the NEC, NFPA 70, governs the electrical installation, the grounding and bonding, and the working clearances in the gray space, while NFPA 110 covers the emergency and standby power where it applies. The fire detection, the suppression, and the life safety follow the building and fire codes and the NFPA suite for the occupancy and the hazard. Across all of it, the project program, the basis of design, and the adopted code edition with local amendments control the specific numbers and the ratios, so verify them against the design and the jurisdiction rather than carrying a remembered figure.
Units, terms, and abbreviations
Data center facility work runs on a stack of terms that get used loosely, so the same word can mean different things on one project. Pin the term to the zone before you act on it.
- White space
- The conditioned data hall where the IT racks and servers sit
- Gray space
- The back-of-house support area for the mechanical and electrical plant
- MMR / meet-me room
- The space where carriers and networks interconnect inside the building
- Entrance facility
- The entrance room where outside cabling enters the building
- CRAC / CRAH
- Computer room air conditioner and air handler, the perimeter cooling units
- CDU
- Coolant distribution unit, the interface for a liquid-cooled row
- Stranded capacity
- Installed power, cooling, or space that cannot be used because another ran out
- Raised floor / slab
- Access floor on pedestals with a plenum, versus racks on the structural slab
- Cage / suite
- A fenced enclosure or a walled room for one colocation customer's gear
FAQ
What is white space in a data center?
White space is the conditioned data hall where the IT racks, servers, storage, and network gear sit, laid out in rows with the power and cooling that feed them. It is the floor the whole building is built to serve, measured by area and, increasingly, by the power and cooling it can deliver to each rack.
What is the difference between white space and gray space?
White space is the data hall where the IT gear runs. Gray space is the back-of-house support area for the mechanical and electrical plant, the UPS, switchgear, generators, and cooling, that keeps the white space alive. One zone computes, the other powers and cools it, and they are walled and access-controlled apart.
What is gray space in a data center?
Gray space is the support area that holds the mechanical and electrical plant, the UPS and batteries, switchgear, transformers, generators, chillers, and pumps. None of it computes; all of it powers and cools the white space. It is sized to the white space it carries, and that size grows as the power per rack climbs.
What is a meet-me room?
A meet-me room (MMR) is the secure space where the carriers and networks in a data center physically interconnect and cross-connect. It lets tenants reach the outside world and lets carriers exchange traffic inside the building. It is a high-security zone, often in fire-rated walls, and larger buildings run two for redundancy.
What is hot aisle and cold aisle layout?
Hot aisle and cold aisle is the row arrangement where cabinet fronts face fronts across a cold aisle and backs face backs across a hot aisle. Cold supply air feeds the intakes, hot exhaust dumps into the hot aisle, and the two are kept from mixing. Containment encloses one aisle to tighten the separation.
What is the white-to-gray space ratio?
The white-to-gray ratio compares the IT floor to the support floor behind it, and it has moved against the white space as density climbed. There is no single correct figure; it depends on density, redundancy, and the cooling approach. Treat any published ratio as a rough planning number and size the gray space from the actual load.
Is a raised floor or a slab better for a data center?
It depends on the density and the cooling. Raised floors use the underfloor plenum to push cold air up through tiles, which suits moderate density. New high-density builds lean toward slab with overhead distribution, because liquid and in-row cooling do not need the plenum and the slab carries the heavier point loads with simpler anchoring.
What limits a data center's capacity?
Capacity is the smallest of three: floor area, power, and cooling. On a modern floor it is almost never the space. Stranded capacity is installed resource you cannot use because another ran out next to it, like floor space with no power left to feed it. Plan all three together and size to power and cooling.
How is a data center zoned for security?
A data center is secured in layered rings: the site fence, the lobby, the general interior, the white space, and the customer cage. Each ring is a separate access decision, enforced at sensitive boundaries by a mantrap, two interlocked doors with a second factor. A colocation tenant reaches only their own cage, never the gray space.
People also ask
Codes cited in this guide
This guide is written and reviewed against the published standards below. Always confirm the current adopted edition with the authority having jurisdiction.