Servers need power to operate. Every millisecond of power interruption results in server downtime. That’s why modern data centers use sophisticated power systems to keep your projects online, even when problems arise with the main power supply.
Let’s look at how these systems make this level of reliability possible. The article ahead will take you through:
- The design of data center power supply systems
- Different ways to achieve power redundancy
- What happens if main power fails
- How backup systems keep your servers running
- Why modern power management matters for your projects
Throughout this article we will be using Hub Europe, our flagship data center, as an example to show you how data center power works not just in theory, but also in practice.
Data Center Power Requirements
Data centers consume massive amounts of electricity – as much as thousands of homes combined. Your servers, along with cooling systems and other infrastructure, need megawatts of power delivered reliably around the clock. To put this in perspective, a 3.1 MW data center like Hub Europe uses roughly the same amount of electricity as a small town.
While a house can survive occasional power cuts without many issues, data centers need to support the workload customer run on the servers located inside..
Reliability Tiers System
The way this continues power is ensured has been codified by the Uptime Institute into a tier-system. Different data centers tiers offer significantly different reliability levels:
Tier I data centers provide basic capacity with no redundancy. They have a single path for power and cooling, offering 99.671% uptime (up to 28.8 hours of downtime yearly). During maintenance, these facilities must shut down completely, making them suitable only for non-critical workloads.
Tier II data centers add some redundant components but still maintain a single distribution path. They offer 99.741% uptime (up to 22 hours of downtime yearly). While they include redundant critical power and cooling components, they lack the multiple distribution paths needed for concurrent maintenance.
Tier III data centers require :
- 72-hour power backup for extended outages
- N+1 redundancy: every critical system has to have at least one additional backup which can take over the whole load of the original system. This allows for maintenance without interrupting your services.
The above-mentioned standards are meant to ensure 99.982% uptime. That’s up to 1.6 hours of potential downtime yearly.
Tier III principles are incorporated in our Hub Europe Data Center by having dual power feeds, multiple backup systems, and independent power paths to provide power all the way to your servers. The system is set up like a network of underground highways carrying electricity through separate, protected tunnels. Even if one “road” gets blocked, electricity keeps flowing smoothly along the alternate route.
Power’s Journey Through the Data Center
Power takes a carefully planned path from utility companies to servers in the data center. Let’s follow this journey through each stage to get a better understanding of the power flow.
From Utility to Facility
Power enters through high-voltage utility feeds – typically 2 to 30+ kilovolts. Similar power requirements can be found in shopping malls, regional hospitals, university campuses, and heavy industrial operations like steel mills or chemical plants.
Modern data centers like Hub Europe use two independent feeds for redundancy. Each 20 kV connection creates its own power chain with dedicated equipment – if one utility connection fails, your servers continue running on the other.
Transformers convert incoming high-voltage electricity to levels the facility can use, stepping it down to 480 volts. This creates two completely independent power paths from the very first stage. By comparison, the household electronics transformer that powers your laptop (typically converting wall outlet voltage to 19 volts) performs the same basic function as these massive data center units, but on a tiny scale – handling watts instead of megawatts.
The complete power chain creates multiple conversion points, each stepping down voltage to more manageable levels:
- Utility voltage (20+ kV coming into the data center)
- Facility distribution (480V after transformer conversion)
- Equipment level (400V in modern facilities or 208V in traditional setups)
- Server-ready power (typically 208V/120V single phase or 400V/230V three-phase depending on server requirements)
Each step reduces voltage to more manageable and safer levels while maintaining the power capacity needed to run server equipment. Modern servers typically have power supplies that accept a range of voltages (100-240V), allowing them to operate efficiently regardless of minor variations in the final power delivery.
Central Distribution
From the transformers, power flows to a central hub (like the Low Voltage Main Distribution – LVMD – system in the case of Hub Europe). This sophisticated system combines both power distribution and transfer switching functions in one central hub. It acts as a traffic controller, directing power through multiple paths:
- Through UPS systems to your servers
- Directly to facility systems like cooling
It also handles extra / alternative power sources – in the case of Hub Europe, for example, 250 kW solar array and an emergency diesel generator. Speaking of solar power at Hub Europe the 250 kW array generates enough electricity to power around 125 houses simultaneously.
The solar array is just one part of the broader sustainability initiatives in place at Hub Europe, which have enabled Contabo to apply for ISO 50001 certification for Hub Europe as a testament to our commitment to sustainability. Find out more on our Sustainability page.
Protection and Backup Systems
Incorporated into the LVMD are Automatic Transfer Switch (ATS) components. These components constantly monitor incoming power quality, watching for:
- Complete power loss (blackouts)
- Voltage drops (brownouts)
- Short-term fluctuations
When utility power fails, a sequence of events keeps servers running:
- UPS systems activate within milliseconds, providing instant backup power
- Emergency generators start immediately, beginning their warm-up sequence
- During the next 10-15 seconds, generators reach full speed and stable output
- The ATS continuously monitors generator power quality
- Once generator power stabilizes, ATS switches servers from UPS to generator power
- When utility power returns, ATS verifies its stability before switching back
- Throughout these transitions, servers continue running without interruption
Generators typically take several seconds to reach full power and stabilize. During this transition period, UPS systems bridge the gap, ensuring servers never notice the switch.
The UPS approach varies by facility – some use battery-based systems providing up to 15 minutes of power, while others employ flywheel systems for shorter coverage. Hub Europe, our example, uses battery-based UPS.
The UPS system itself is redundant for additional protection and impact-free maintenance. Dual UPS paths create two completely independent power protection systems for your servers. At Hub Europe, this means your equipment receives power from two separate UPS systems simultaneously, each capable of handling the full load if the other fails.
When one UPS needs maintenance, servers continue running on the second path without interruption. This dual-path approach eliminates single points of failure in power protection system – if one UPS experiences issues with batteries, electronics, or needs firmware updates, the redundant UPS keeps everything running smoothly.
Final Delivery
Floor-mounted Power Distribution Units (PDUs – devices that distribute electrical power to multiple pieces of equipment in a data center) perform the next voltage conversion. They step 480V power down to either 400V (in modern facilities) or 208V (in traditional setups). The 400V option reduces power loss during distribution, improving overall efficiency.
From PDUs, Remote Power Panels (RPPs) distribute power through dedicated circuits, each protected by its own breaker. While traditional facilities run power through under-floor cables, some newer data centers use overhead bus systems for more flexible power delivery.
At each rack, power whips terminate in standardized outlet boxes mounted at the top or bottom of the server cabinet. These boxes connect to two separate rack PDUs – one from each power chain. Modern servers take advantage of this dual-power setup through redundant power supplies. Each power supply connects to a different PDU, drawing power from separate chains. If either chain experiences issues, the server automatically runs entirely on the remaining supply without interruption.
Redundancy in Action
As we’ve seen, modern data centers use multiple layers of protection to maintain power in case problems arise. To illustrate this, let’s take a closer look at how things work at Hub Europe when different scenarios arise.
Normal Operations
During standard operation, power flows from the primary 20 kV delivery point through the power feed-in and transformer system, as well as the photovoltaic array. The LVMD receives 3.1 MW of power from the transformer (supplemented by up to 250 kW from solar) and distributes it across three paths:
- Through Feed 1 to the UPS system
- Through Feed 2 as a redundant UPS path
- Directly to facility systems like cooling and lighting
Primary Power Failure
If the main power delivery point fails, the backup 20 kV connection activates automatically. The transition happens seamlessly through the power feed-in system, maintaining consistent power delivery to all systems.
Complete Utility Failure
In the rare situation when both utility power sources would fail, multiple systems engage:
- The UPS system provides immediate power to critical loads
- The 3.1 MW emergency generator activates
- The bypass circuit allows direct generator connection if needed
- Solar power continues contributing up to 250 kW to the LVMD if possible
Maintaining Without Interruption
Aside from power outage scenarios, data centers require regular maintenance to ensure reliable operation. This includes UPS battery replacements, generator testing and fuel management, power distribution component inspections, and infrastructure upgrades. Hub Europe’s redundant design means these essential activities happen without affecting servers. For example, during UPS maintenance:
- Engineers first verify the alternate power path can handle the full load
- They activate bypass circuits to isolate the maintenance area
- Power flows through redundant systems while work proceeds
- After testing, systems reconnect without any service interruption
As mentioned earlier, the N+1 redundancy standard means every critical system has at least one independent backup. This redundancy extends through all protection systems – from power feeds to UPS units to distribution paths. If any single component needs maintenance or experiences a failure, your applications remain protected by the backup systems.
Real-World Testing
System testing is used to prove how well these protection layers work together. For example, stress testing at Hub Europe, conducted in August 2024, involved gradually increasing loads across different power paths while monitoring all systems. Even at peak loads of 1,800 kW, voltage remained stable and backup systems maintained readiness.
The Bottom Line
When you host your applications in a modern data center, multiple power protection systems work together to keep your projects online. From dual utility feeds to rack-level redundancy, each component adds another layer of protection for your servers.
At Hub Europe, this sophisticated power infrastructure proves its worth every day. With comprehensive backup systems, your servers get the reliable power they need to run without interruption.