Liquid Web Data center power? The numbers are staggering — data centers consumed around 460 terawatt-hours (TWh) of electricity in 2022, more than the national energy consumption of some countries for the same year. And there are projections that this number will rise to 1,000 TWh in 2026.
Downtime from power failures can wreck your business. Frustrated customers and lost revenue aren’t a pretty picture. Then there’s the efficiency question. Data centers guzzle electricity, generating killer bills and a massive carbon footprint.
Get it right, and you have reliability, efficiency, and peace of mind.
This article covers everything about data center power, including why it’s important and how it supplies and distributes power within it.
- What is data center power?
- Why data centers power matters
- What are data centers powered by?
- How is power supplied to data centers?
- Six ways power is distributed in a data center
- Power system redundancy methods in data centers
- Green data center initiatives
- Server rack power consumption
- Utility power capacity vs. IT power capacity
- FAQ on data center power
- Power your data center with Liquid Web
What is data center power?
Data center power is the infrastructural architecture that works around the clock to provide power to a data center, manage it, and protect it. This process includes power generation, distribution, and management within the data center.
Proper data center power infrastructure ensures the continuous operation of IT equipment, cooling systems, and other components within the data center. It’s essential in maintaining uptime, supporting high-density computing environments, and enabling energy efficiency and cost optimization in data center operations.
10 essential data center power metrics and KPIs
The efficient running of a data center power infrastructure relies on certain key metrics, which you should consider.
1. Power usage effectiveness (PUE)
This compares the total power used by the data center with the total power supplied. Think of it as a report card for energy efficiency. Aim for a goal of PUE of 1.2 or less.
2. Data center infrastructure efficiency (DCIE)
This is the ratio of the energy consumption of IT equipment in the data center with the data center’s total power consumption.
3. Rack cooling index (RCI)
It measures the data center racks’ cooling rate effectiveness compared to the industry standards of rack temperatures.
4. IT power consumption
This measures the power consumption by data center components that power IT equipment such as servers.
5. Cooling power consumption
This is the power consumption of the data center’s cooling components, such as HVAC systems.
6. Power density
This measures the total power supplied to a rack or designated area in a data center. It’s measured in Kilowatts.
7. Carbon usage effectiveness (CUE)
CUE tracks the carbon released per hour after every kilowatt of energy a data center uses.
8. UPS efficiency
Your uninterruptible power supply is insurance against outages — it involves storing power in batteries as a backup for emergencies, such as power outages in the data center. This metric evaluates whether it’s running optimally.
9. Incident response time
When things go sideways in your data center, you need to know how quickly your team can get the power back on.
10. Renewable energy percentage
Reduce your carbon footprint by maximizing your renewable energy mix. This metric shows your progress.
Why data center power matters
When the power goes out, your data center operations stop. Optimizing your power usage from the core to the fringes is an endless battle you shouldn’t ignore.
Servers and IT equipment
In a functional data center, servers and IT workhorses are major components that ensure the continuous operation of the data center. Therefore, it’s essential to provide a constant flow of electricity to supply power to the servers and IT equipment, such as CPUs, fans, hard drives, and others.
Cooling systems
With all that computing power in action, proper cooling is essential. Your cooling setup should include systems like HVAC to maintain the data center at industry-standard working temperatures.
Lighting and other infrastructure
Secondary power consumers in a data center, such as lighting, security systems, and monitoring gear, also require electricity. This includes lighting up the aisles for equipment inspections and powering security systems like alarms and cameras for maximum protection.
What are data centers powered by?
Data centers require a reliable power source to operate. Let’s examine the power sources used.
Alternating current in data centers
Data centers use two types of electrical current: alternating current (AC) and direct current (DC).
AC power is commonly used in data centers as it integrates with existing electrical grids and infrastructure. It enables long-distance power distribution and voltage adjustments using transformers, making it the standard choice for most data center operations.
Direct current in data centers
DC power is stored in batteries or UPS systems, providing backup power during outages. Due to its efficiency benefits, it is gaining adoption, particularly in large hyperscale data centers. It delivers power directly with minimal conversion losses, resulting in significant cost savings for high-energy-consuming facilities.
What voltage is data center power?
Data centers utilize different voltage levels for specific applications. Low voltages between 120V and 240V power IT equipment like servers and networking gear. Higher voltages of 480V and above distribute power across facilities, minimizing energy losses over distances.
Single-phase vs. three-phase power
Voltage is just one part of the power puzzle. Data centers also need to choose between single-phase or three-phase alternating current.
Single-phase power supplies a single oscillating wave, suiting smaller loads. It powers residential settings or essential equipment. Three-phase power supplies a constant flow by rapidly alternating multiple oscillating waves. It efficiently meets the high power demands of large data centers.
| Aspect | Single-phase power | Three-phase power |
| Power supply | Single oscillating wave | Multiple alternating oscillating waves |
| Typical use | Residential settings, essential equipment | Large data center operation |
| System complexity | Simple | More complex |
| Installation costs | Lower | Higher |
| Capacity | Limited | Higher voltage capacity |
| Efficiency | Lower | Higher, with reduced energy losses |
| Equipment costs | Lower | Higher, requires specialized equipment |
How is power supplied to data centers?
The power supply journey to a data center involves multiple stages and systems working together. Let’s trace that path from its origins.
Generation
Power generation starts at various plants, including coal-fired, nuclear, hydroelectric, wind, or solar farms. These generating stations produce the initial electrical supply. The electricity, typically generated at a low voltage of about 11.5–16kV, is stepped up to 330kV by transformers for transmission.
Transmission
Power is transmitted from the plants over long distances through high-voltage transmission lines. The 330kV electricity is transported to substations and stepped down to 132kV. It’s further transmitted to injection stations and stepped down to 33kV.
Distribution
The high voltages are stepped down at the substations for distribution over local utility lines. Distribution transformers step down the voltage from 33kV to 11kV and then to 0.415kV, finally reaching 240V to deliver electricity directly to businesses and data centers.
Conversion and conditioning
Final conversions and conditioning are done within the data center to ensure each component gets the exact voltage and current needed. Power conditioning equipment, like filters, voltage regulators, and uninterruptible power supplies (UPS), helps deliver a clean, stable supply by mitigating issues like spikes, harmonics, and outages. This ensures a reliable, high-quality power flow to maintain continuous data center operations.
Six key ways power is distributed in a data center
Once power is supplied to the data center, it needs to be distributed to all components. The distribution points are as follows.
Utility grid
The primary power source to the data center is the utility grid, which consists of the production, transmission, and distribution grids that make power available.
Transformers
Transformers receive high-voltage electricity from the distribution grids and step down the voltage to a safe and efficient level for the data center.
Switchgear
Think of switchgear as the traffic control center for power distribution. It involves constant monitoring and includes various devices that control, regulate, and protect the power supply in the data center.
PDUs and UPS
Power distribution units (PDUs) and the uninterruptible power supply (UPS) work together to facilitate energy flow within the data center. The UPS stores energy for emergencies, while the PDUs convert this energy to suitable currents and voltages for distribution to data center components.
Backup generators
When utility power fails for an extended period, diesel or gas-powered generators kick in to supply electricity and prevent downtime. Backup generators serve as external or secondary power sources for the data center.
Redundant power paths
Redundant power paths provide multiple power transmission lines to the data center to ensure a reliable electricity supply. In case of a transmission line failure, other power paths are activated immediately.
Power system redundancy methods in data centers
Regardless of the robustness of a data center’s power supply, the possibility of a power failure always looms. Therefore, the more redundancy you build, the more resilient your power supply becomes in an emergency.
The N+1 model provides a straightforward approach to redundancy. “N” represents the number of components necessary to run the system under normal operating conditions. The “+1” indicates the presence of one additional backup component.
N redundancy
While N doesn’t offer backup power during an emergency outage, it reliably meets a data center’s minimum power requirements to ensure its continuous operation.
N+1 or N+X redundancy
This redundancy system is designed to provide a single or an extra number of backups for a data center in case of power failure. It ensures that a data center keeps running until the main grid restores the power supply.
2N redundancy
2N redundancy is not just a backup; it’s a mirror image of the minimum power needed to run a data center. This redundancy system efficiently supplies the exact amount of power required by the data center — and doesn’t necessarily need to be equal to capacity.
2N+1 redundancy
Similar to 2N redundancy, 2N+1 redundancy duplicates a data center’s power requirements with an additional backup. This is an extra two layers of power supply for a data center. For example, if a data center requires 100 power units, 2N+1 redundancy will provide 200, with an additional 100 units as backup.
3N/2 redundancy
3N/2 redundancy, or three-to-make-two redundancy, provides an extra power supply to the data center components but not up to double the required power. For example, it provides the power required to power three servers to two servers, leaving the server component with extra power in case of a failure.
The remaining power source will power the data center until power is restored. This ensures continuous operation even if one power source fails without requiring an excessive power supply.
Green data center initiatives
Maximizing energy efficiency is a top priority for data centers aiming to reduce operating costs and environmental impact. In line with this objective, data centers are increasingly adopting green data center initiatives.
Renewable energy integration
Integrating renewable energy sources, such as solar plants and wind turbines, into the power supply chain allows data centers to reduce carbon emissions and environmental impact.
Water usage optimization
Water is often wasted through leaks, inefficient appliances, and overuse. Optimizing your water usage means taking smarter actions about how you consume this resource. That could mean fixing dripping faucets, installing low-flow showerheads, or landscaping with drought-tolerant plants.
Green certifications and standards
Businesses and buildings can go the extra mile for sustainability by earning green certifications like LEED or Energy Star. These are rigorous standards for energy efficiency, water conservation, indoor air quality, and more. Getting certified shows a commitment to reducing environmental footprints through smart design and operating practices.
Energy-saving technology
Using low-power processors reduces energy consumption in data centers. Server virtualization also improves energy efficiency.
Server rack power consumption
Depending on their power density, typical server racks consume energy somewhere in the 5kW to 30kW range. However, as modern racks continue to increase in density, they may need up to 200kW of power. Artificial intelligence and big data center servers are expected to consume more energy, constituting higher-density power racks.
Utility power capacity vs IT power capacity
Utility power capacity refers to the maximum power the data center can provide at any given time. This includes all the energy the data center consumes, including the IT equipment.
On the other hand, IT power capacity refers to the maximum power that can be consumed by IT equipment at any given time.
It’s important to maintain a balance between the utility power capacity and the IT power capacity. The utility power capacity should be higher than the IT power capacity. This ensures that the IT equipment has enough power to operate properly, preventing disruptions in the data center’s operations.
FAQs on data center power
Learning about data center power systems can raise questions. Here are some common ones.
What are the power supply requirements for a data center?
The power supply requirements depend on the data center size. A small data center with around 2,000 servers typically requires about 5MW, while hyperscale data centers with tens of thousands of servers may need up to 100 MW of power.
Which components consume the most power in a data center?
Servers consume the most power in a data center due to their processing requirements and cooling needs. The cooling system also often requires a large portion of a data center’s total energy consumption for optimal temperature and humidity.
Other components, like lighting, physical security, and fire suppression systems, also consume power. However, their contribution to total energy consumption is typically lower than that of servers and cooling systems.
Power your data center with Liquid Web
The power systems within data centers are vital to ensure continuous and reliable operations. From sophisticated redundancy methods to the types of current and voltage used, each component is essential to maintain uptime and efficiency.
Understanding the power flow, distribution hierarchy, and challenges data center operators face provides insight into modern data centers‘ complexities. As the demand for data processing grows, advancements in power technology will continue to drive the evolution of these critical infrastructures.
Partner with Liquid Web to take advantage of excellent data center solutions and other IT solutions — take your business’s technology strategy to the next level.