Managing a Server and Communications Room in a Building — Three Critical Systems the Building Manager Must Know

In every modern office building there is one small room where a half-hour outage can paralyze dozens of businesses at once — the server and communications room. Most building managers treat it like a storeroom with an air-conditioner, and that is exactly where the problem begins. A data-center room is not just another room; it is a critical system in which three sub-systems — precision cooling, backup power, and dedicated fire detection and suppression — must work together, continuously, 24 hours a day. When one of them falls, the others are not enough to save the situation.

I have managed buildings with data-center rooms, and I have seen the same pattern repeat: the manager knows the machine room, the roof, the main electrical panel — but the little room with the green flashing lights he leaves to "the computer company." The problem is that the computer company takes care of the servers themselves, not the physical infrastructure they sit inside. Responsibility for the cooling, the power and fire prevention lies with the building manager — even if he doesn't know the manufacturers' names.

Why a Data-Center Room Is Different from Any Other Room in the Building

An ordinary office room tolerates heat, humidity and a power outage of a few minutes without damage. A data-center room does not. The equipment inside it emits heat continuously, does not stop working even for a moment, and is sensitive to temperature, humidity, dust and power interruptions. Three characteristics make it critical:

• Concentrated heat load: in a small area — sometimes less than ten square meters — switches, servers, routers and storage arrays are crowded together, all of them turning electricity into heat. Without continuous heat removal, the temperature climbs within minutes, not hours. I saw a room that reached 45 degrees within twenty minutes of the cooling failing.
• Zero tolerance for interruption: a power drop of a single second brings down unprotected servers. A prolonged outage causes data loss and downtime for everyone who depends on the room — residents, access systems, cameras, and sometimes even the building's own safety systems.
• Hidden fire risk: a high concentration of electrical equipment, dense cabling and many connection points create a fire risk that is not visible to the eye until it's too late. A water sprinkler above such equipment is a disaster in its own right.

The practical conclusion: this room needs a separate management regime, not applying to it the same maintenance routine as a corridor or a meeting room. This is also why in high-tech buildings and large business complexes these rooms are managed as an asset in their own right, as I expanded in high-tech campus management.

Precision Cooling — The Heart That Must Not Stop Beating

The most common mistake is to install an ordinary overhead air-conditioner in a data-center room, of the same type that cools a meeting room. A comfort air-conditioner is designed to cool people intermittently — not equipment that emits heat continuously for many hours. In practice, many buildings get by for years with a domestic air-conditioner, and then comes a summer day with a peak load and 45 degrees outside — and it simply cannot remove the required quantity.

A data-center room needs precision cooling — dedicated units that maintain not only temperature but also stable humidity (usually 45%–55%), filter dust, and are designed to operate 24/7 for years. Humidity that is too low creates static electricity that harms electronic components; humidity that is too high produces condensation on electronic boards.

What Is Maintained in the Cooling and Why It Fails

• Filters: a clogged filter reduces airflow, gradually raises the temperature and forces the condenser to work harder. Cleaning or periodic replacement is the cheapest action with the greatest impact — and in a data-center room you do it three times as often as in a regular room.
• Refrigerant gas and pressures: a slow gas leak lowers cooling capacity long before the unit shuts down entirely. That is why you check pressures — not just "whether it's cold."
• Condensate drainage: a clogged drain causes a silent flood beneath the raised floor or onto the equipment itself. This is one of the failures discovered too late, usually after damage has already occurred.
• Backup and redundancy: in a critical room you do not settle for a single cooling unit. The N+1 principle means there is always one unit beyond what is required, so that a failure of one unit does not bring the room down.
• Temperature and humidity monitoring: sensors that alert on a deviation before it becomes damage — preferably linked to a building management system (BMS) so that the alert arrives even outside working hours.

Precision cooling is a sub-field of air-conditioning, and the same preventive maintenance principles apply to it. I expanded on the maintenance of air-conditioning systems in a building in air-conditioning maintenance in office buildings — the difference in a data-center room is mainly the frequency, the level of redundancy and the fact that there is no "quiet season" in which you can neglect it.

Backup Power — Two Layers of Protection That Must Work Together

Backup power for a data-center room is not a single component but a chain of two layers whose roles are completely different, and many confuse them. Understanding the difference is critical, because each layer requires different maintenance — and one cannot replace the other.

First Layer: UPS

The UPS kicks in at the first moment of a power drop, without any delay, and supplies electricity from its batteries. Its role is not to hold the room for hours but to bridge those seconds until the generator comes up — and also to filter the quality of the power from tiny spikes and dips in the grid, which also harm sensitive equipment.

The batteries are the classic wear part: they wear out with time and with heat, and a UPS with a dead battery is a UPS that will fail exactly at the moment of truth. I saw cases where a UPS displayed "normal status" on the dashboard but in practice the batteries were no longer usable — because they were never tested under load. That is why you test them periodically and don't wait for a failure alert.

Second Layer: Generator

The generator supplies electricity during prolonged outages, after the UPS has bridged the first seconds. It requires a completely separate maintenance regime: periodic ignition tests under load (not just idle starting), checking the fuel level and its quality, the starter battery, and the automatic transfer switch (ATS) that times the changeover.

A generator started once a year "to see that it works" is no guarantee of anything — the failure occurs precisely under a real load. The common pitfall in Israel: the generator test is carried out in the rainy half of the year, when the load is low; it fails precisely in summer, when the cooling adds a heavy load and the run duration is long.

These two layers are part of the building's overall electrical setup, and the same safety and grounding principles apply to them. I expanded on this in electrical systems maintenance in an office building. In a data-center room there is simply no margin for error: every link in the chain must be tested, because the weakest link is the one that will determine the outcome.

Dedicated Fire Detection and Suppression — Why a Water Sprinkler Is a Disaster Here

In a data-center room, the building's standard fire suppression system can do no less harm than the fire itself. A water sprinkler that sprays over active electrical equipment causes immediate, irreversible damage, even if the fire is small and localized. That is why a critical room requires a dedicated response on two fronts: early detection and suppression that does not harm the equipment.

Early Fire Detection

Electronic equipment tends to "overheat" and emit thin smoke long before an open fire ignites. Air-sampling systems (VESDA or a similar trade name) continuously sample the air and detect smoke particles at a very low concentration — usually before an ordinary smoke detector would have reacted. This is the difference between replacing a single component and losing the entire room. I saw a case in which a short-circuited wire was replaced because the system detected a light burnt smell — the servers weren't even affected.

Gas Suppression Instead of Water

In critical rooms, clean-gas suppression systems are used (for example FM-200, Novec 1230 or CO₂) — a gas that extinguishes the fire without leaving residue and without harming the electrical equipment. The gas requires the room to be sufficiently sealed to hold the suppression concentration for the required period, and therefore the sealing of the room itself is an integral part of the system. Every hole in a wall, a ceiling or a cable duct is a potential failure.

Maintenance here includes: checking the cylinder pressure, the integrity of the actuation heads, signage and clarity of escape routes, and testing the room's sealing — usually in a Door Fan test per standard.

• Separation from the building system: the dedicated suppression system must be coordinated with the building's main detection system, but operate on its own logic and not activate sprinklers.
• Human preparedness: releasing gas requires the room to be evacuated before activation, and therefore there is an activation delay (usually 30–60 seconds) and an audible and visual alert — which must be verified as working and audible.
• Automatic power disconnection: upon activation of the suppression system, the power feeds to the equipment in the room must be disconnected — to prevent re-ignition and preserve the integrity of the gas.
• Documentation with the Fire and Rescue Authority: a dedicated suppression system is part of the building's safety file and requires approval and periodic inspection with the authority.

Fire detection and suppression obligations for the entire building, including the obligations with the authorities, are detailed in the fire safety law and rules for an office building — and the critical room is an exception that tightens the general requirements, not eases them.

The Three Systems Are One System — The Hidden Point of Failure

The deepest managerial mistake is to maintain the three systems separately, as if there is no connection between them. In practice they are interlocked, and a failure in one brings down the others in a chain:

• Cooling depends on power: in a power outage, even if the UPS holds the servers — the cooling may shut down if it wasn't designed for it. Within minutes the temperature will climb until an automatic emergency shutdown of the servers. That is why you design the power backup to feed the cooling too, not just the servers.
• An electrical failure = a source of fire: an electrical short, overload on old connectors, or wear of unmaintained power lines are among the common causes of fire in data-center rooms. The suppression system must disconnect power feeds upon activation.
• Heat is also a fire risk: a cooling failure raises the temperature, and a continuous heat load accelerates wear and electrical faults — which are exactly the source of fire.

The tool that allows you to see the three systems as one picture is a building management system (BMS) that consolidates sensors, alerts and response logic in one place. I explained this in depth in the guide to building management systems. In a data-center room, a BMS is not a luxury — it is your eyes when you are not in the room at two in the morning.

Regular Inspections: A Practical Checklist for the Building Manager

The following list is based on experience working with data-center rooms in office buildings, combined with manufacturer requirements and maintenance standards. It does not replace a certified engineer, but it is what the building manager needs to verify is being done:

Ongoing Monitoring (Daily and Automatic)

• Automatic monitoring of temperature and humidity with alerts to SMS or email — not a weekly manual check.
• Checking the UPS indicator panel: charge status, battery status, warnings.
• A quick visual check: green lights, no burnt smells, the door closed.

Periodic Maintenance (Monthly to Quarterly)

• Cleaning and replacing cooling filters — at a higher frequency than regular rooms.
• Checking condensate drainage and duct cleanliness.
• Checking refrigerant gas pressures in the precision cooling units.
• Checking the UPS batteries — pressure, capacity, and expiry dates.
• Running the generator under load, not just idle starting.

Annual and Cyclical Inspections

• Checking the gas suppression system: cylinder pressures, actuation heads, actuation and delay conductors.
• Checking the room's sealing (Door Fan Test) — per the suppression system's instructions.
• Checking the early smoke detection system — cleaning the sampling heads, verifying sensitivity.
• An electrical cabling survey — wear, loads, undocumented additions.
• Checking the ATS (automatic transfer switch) under a simulated outage condition.
• Updating the safety file and coordination with the Fire and Rescue Authority (required under the Business Licensing Law and the fire regulations).

Documentation and Control

• An orderly maintenance log: every inspection, finding and fault — with a date, the technician's name and a signature.
• Controlled access management: only authorized parties enter, with documentation of every entry.
• Cleanliness and airing: a room clean of dust, without storage of waste, envelopes or flammable materials.

All these actions should be integrated into the building's overall annual cycle. I built a full template in the annual preventive maintenance checklist — here you simply tighten the frequency and the level of redundancy. The documentation principles and the framework are drawn, among others, from Israeli Standard SI 1525 for building maintenance.

The Economic, Legal and Liability Side

In Israeli building culture, people tend to act only when something external forces it — a regulator, an inspection or a fault that has already happened. In a data-center room this approach is especially expensive, because the neglect does not manifest as a stain on the wall but as downtime that paralyzes whole businesses at once.

A failure of the data-center room harms all the tenants who depend on it. As the building manager you are the one who will be called to explain why the generator did not come up, why the cooling failed, and where the inspection records are. An orderly, documented maintenance log is your legal and insurance protection — when something goes wrong, it is the proof that you did what was required.

Beyond liability, there is also a simple economic logic here: the cost of replacing UPS batteries, an annual service for the generator and checking gas cylinders is a small fraction of the cost of a building's downtime for several days, equipment damage, and tenant claims. A properly maintained property also preserves its value and the systems' lifespan, instead of being eroded by outages and recurring faults.

At Domera we manage server and communications rooms as a critical asset in their own right — a separate maintenance program, continuous monitoring and a documented log with insurance and the authorities. For details on the service: property management, or get a quote.