We have all seen it at some point. You open a door expecting a broom cupboard and find a 42U rack sitting between a mop, a yellow bucket and three bottles of bleach. The LEDs blink with quiet resignation, the server fan sounds like an industrial hair dryer and the thermometer reads 38 degrees Celsius. Someone, at some point, decided this was a reasonable place to house the company's critical infrastructure. Spoiler: it was not.
This article is for you if you have ever thought that a rack in a random room is "basically the same" as a data centre. It is not. Not even close. And we are going to explain why with data, humour and a comparison table that should be printed and taped to the door of every cleaning cupboard in the world.
The Classics: Where a Server Should Never Be
Before diving into the technical detail, let us take a tour of the "best" places where we have found production servers over the years. This is not fiction: every single one of these examples is real.
- mop The cleaning cupboard: the undisputed classic. A rack sharing space with chemical products, residual mopping moisture and ventilation that consists of "leaving the door slightly open". Bonus: the cleaning staff unplug the server to plug in the vacuum cleaner.
- stairs Under the stairs: Harry Potter had a bedroom there. Your server should not. Dust, vibrations from footsteps and a temperature that rises every time someone runs up to the second floor.
- restaurant Next to the kitchen: airborne grease, steam, ambient temperature above 30 degrees in summer. The server doubles as an extra heater for the office and nobody understands why the ERP slows down after lunch.
- person In the CEO's office: "this way I can keep an eye on it," says the director while the server generates 65 dB of constant noise and turns the office into a tropical greenhouse. Access control: the office door.
- wc In the bathroom: yes, the bathroom. The logic was that "nobody steals anything from the bathroom" and that "there is always ventilation". The logic was wrong on both counts. The constant humidity destroyed the disks in under a year.
If you recognised yourself in any of these scenarios, do not worry: you are not alone. But keep reading, because what follows should convince you that there is a better way of doing things.
What Makes a Real Data Centre
A real data centre is not simply a place to put servers. It is a facility designed, built and operated specifically to keep IT equipment running continuously, securely and efficiently. Every aspect of the design is engineered to eliminate risk:
- bolt Power redundancy (N+1, 2N): dual independent utility feeds, automatic transfer switches (ATS), UPS systems to cover micro-second switchover and diesel generators with hours or days of autonomy. If a power line fails, the datacenter does not even notice.
- ac_unit Precision cooling: CRAC/CRAH systems that maintain constant temperature (18-27 C) and humidity (40-60%). Separated hot and cold aisles, raised floors for controlled airflow, sensors in every rack. None of that "open the window when it gets hot" approach.
- local_fire_department Fire suppression: early detection systems (VESDA) and inert gas suppression (FM-200, NOVEC 1230) that extinguish fires without damaging equipment. An ABC powder extinguisher would destroy the servers just as thoroughly as the fire itself.
- security Physical security: biometric access control, 24/7 CCTV, entry/exit logging, visitor escorts, reinforced physical perimeter. A door with a key that everyone has is not enough.
- monitoring 24/7 monitoring: sensors for temperature, humidity, vibration, smoke, water and the status of every critical component. A Network Operations Centre (NOC) with permanent technical staff. Any anomaly is detected and acted upon before it becomes a problem.
Your Room vs a Real Data Centre
Let us put things in perspective with a direct comparison. If your "server room" looks more like the left column, you have a problem:
| Criterion | Your room | Real DC (Tier III+) |
|---|---|---|
| Power supply | A single wall outlet, no UPS | Dual utility feeds, UPS + diesel generator |
| Cooling | Domestic air conditioning (or nothing) | Precision CRAC/CRAH, hot/cold aisle |
| Fire suppression | Powder extinguisher in the hallway | VESDA + FM-200/NOVEC, gas suppression |
| Access control | Key that half the building has | Biometric + CCTV + access logging |
| Monitoring | Someone checks "once in a while" | 24/7 NOC, sensors in every rack |
| Availability SLA | None | 99.98% or higher (Tier III) |
| Network redundancy | A single fibre line | Multiple carriers, diversified routes |
| Water protection | Pipes running above the rack | Water sensors, no pipes above equipment |
| Dust and particles | Cleaned when someone remembers | HEPA filtration, positive pressure |
| Insurance coverage | Insurance won't cover uncertified facilities | Full certification, guaranteed coverage |
Real Risks of an Improvised Server Room
This stops being funny when you calculate the real impact of a failure. We are not talking about hypothetical scenarios: every one of these risks generates real incidents every week at companies around the world.
- thermostat Overheating: without precision cooling, the temperature climbs until servers engage thermal throttling (reducing performance to avoid burning out) or simply shut down via thermal protection. A server that shuts down due to heat can corrupt data, file systems and databases.
- power_off Power outages: without a UPS or generator, a power cut shuts everything down instantly. Pending disk writes are lost, RAID arrays can enter a degraded state and databases can become corrupted. And when the power returns, the startup process can take hours if a rebuild is needed.
- water_damage Water damage: pipes running above the rack, leaks from the floor above, overflow from the adjacent bathroom. Water and electronics are incompatible. A single drip can destroy a server motherboard worth thousands of euros.
- blur_on Dust accumulation: dust clogs heatsinks and fans, raises internal temperatures and accelerates component wear. In a room without air filtration, a server accumulates in months the amount of dirt it would not see in years inside a proper data centre.
- no_encryption Unauthorised access: if anyone can walk into the room, anyone can unplug a cable, reboot a server, insert a malicious USB drive or simply walk off with a hard disk. Without access control, there is no security.
- gavel Insurance problems: most business insurance policies do not cover damage to IT equipment housed in uncertified facilities. If your server floods or catches fire in a cleaning cupboard, the insurer can reject the claim.
The Real Cost of "Saving Money"
The reason many companies stuff a server into a cupboard is to "save money". But that saving is illusory when you calculate the real cost of downtime. According to Gartner data and industry studies, the average cost of downtime varies drastically depending on company size:
- store SME (10-50 employees): between EUR 1,000 and 5,000 per hour of downtime, including lost productivity, missed sales and recovery costs.
- business Mid-size company (50-250 employees): between EUR 10,000 and 50,000 per hour. A half-day outage can easily exceed the annual cost of a colocation service.
- corporate_fare Large enterprise: hundreds of thousands of euros per hour. At this scale, nobody has servers in a cleaning cupboard (we hope).
Key fact:
A single 8-hour downtime incident at an SME can cost more than a full year of colocation in a professional data centre. The "saving" from not paying for hosting evaporates with the first serious outage.
Add to this the compliance issues. If your company handles personal data (and every company does under the GDPR), hosting servers in facilities without adequate security measures can constitute a breach of Article 32 of the GDPR, which requires appropriate technical and organisational measures to protect personal data. Fines can reach 4% of annual turnover.
Tier Classification: What Level Do You Need
The Uptime Institute defines four certification levels (Tiers) for data centres. Each level establishes requirements for redundancy, availability and maintainability:
- looks_one Tier I (Basic): single capacity, no redundancy. A single power and cooling path. 99.671% availability (28.8 hours of annual downtime). Requires full shutdown for maintenance.
- looks_two Tier II (Redundant components): adds N+1 redundant components (UPS, generators) but retains a single distribution path. 99.741% availability (22 hours of annual downtime).
- looks_3 Tier III (Concurrently maintainable): dual power and cooling paths, one active and one on standby. Allows maintenance without shutting down any equipment. 99.982% availability (1.6 hours of annual downtime). This is the most common level for serious commercial datacenters.
- looks_4 Tier IV (Fault tolerant): dual active-active paths for everything: power, cooling, network. Withstands any component failure with no service impact. 99.995% availability (26 minutes of annual downtime). Reserved for the most demanding critical infrastructure.
Your cleaning cupboard does not fit into any of these levels. Not even Tier I, which already requires dedicated power and cooling infrastructure.
When a Local Rack Makes Sense
Let us be fair: there are scenarios where having a rack in the office makes sense. A small office with non-critical workloads (a NAS for shared files, a test server, a local domain controller) can justify a local rack. But even then, you need certain basics:
- check_circle UPS: at least 15-30 minutes of autonomy for a clean shutdown or to cover micro-outages. Non-negotiable.
- check_circle Basic cooling: a dedicated air conditioning unit or at least forced ventilation that keeps the temperature below 27 degrees Celsius.
- check_circle Fire detection: at a minimum, a smoke detector connected to the building's general alarm system.
- check_circle Access control: a restricted-key lock, a record of who enters and when. None of that "the door is always open" approach.
If your workload is business-critical (ERP, email, public website, customer databases), a local rack is not the right solution, regardless of the measures you take. The advantages of a European datacenter far outweigh any improvisation.
Alternative 1: Colocation
If you already own hardware (servers, switches, firewalls) and do not want to get rid of it, the most straightforward solution is colocation: you move your rack to a real datacenter and benefit from all its infrastructure (redundant power, cooling, security, connectivity) for a monthly fee.
With colocation you retain full control of your hardware and software but eliminate all physical risks. Your server no longer shares space with mops: it sits in a locked rack, in a climate-controlled aisle, with dual power feeds and a contractually guaranteed availability SLA. You can see how this option compares with the public cloud in our article on colocation vs public cloud.
EasyDataHost offers colocation in its Tier III+ datacenter in Madrid, with multi-carrier connectivity, 24/7 access and remote technical support included. From half a rack to private rooms.
Alternative 2: Cloud and VPS
If you do not need your own hardware (or do not want to deal with it), the cleanest alternative is to eliminate the rack entirely. A VPS server or a cloud IaaS infrastructure gives you the same functionality (or more) without having to worry about hardware, cooling, power or physical security.
With cloud/VPS you pay for the resources you use, scale on demand, get automatic backups and benefit from the entire datacenter infrastructure without managing anything physically. For most SMEs, this is the option with the best cost-benefit ratio: no upfront hardware investment, no physical maintenance, no power worries and an SLA that your cleaning cupboard will never be able to offer.
EasyDataHost: A Real Data Centre in Madrid
EasyDataHost operates from a Tier III+ datacenter in Madrid with all the certifications and guarantees that your cleaning cupboard does not have (and never will):
- check_circle Dual utility power feeds with redundant UPS and diesel generators with extended autonomy.
- check_circle N+1 cooling with precision systems and separated hot/cold aisles.
- check_circle Inert gas suppression (FM-200) with VESDA early detection.
- check_circle 24/7 security: biometrics, CCTV, perimeter access control and permanent NOC.
- check_circle Multi-carrier connectivity with access to the main Spanish and European traffic exchange points.
- check_circle Managed services including monitoring, maintenance and technical support.
Conclusion
A server rack in a cleaning room is not a data centre. It is not, it will not be, and no amount of duct tape or USB fans will change that. Your servers deserve better than sharing space with a mop, and your business deserves infrastructure that does not depend on nobody unplugging anything to plug in the vacuum cleaner.
- arrow_right A real data centre provides power redundancy, precision cooling, fire suppression, physical security and 24/7 monitoring.
- arrow_right The cost of downtime quickly exceeds any "savings" from not paying for professional hosting.
- arrow_right The Uptime Institute's Tier classification defines four levels of redundancy and availability.
- arrow_right Colocation lets you move your hardware to a real data centre without giving up control.
- arrow_right Cloud/VPS eliminates the hardware entirely: pay for what you use, scale on demand.
If your server is still living among mops and bleach, it is time to give it a proper home. Contact our team and we will help you find the solution that best fits your business: colocation, cloud, VPS or whatever you need. But please, get the server out of the cleaning room.