Data Center Energy Consumption

Data center energy consumption refers to the amount of power used on an annual basis, measured in kilowatt-hours (kWh), to power the facility. Data centers are one of the most energy-intensive building types as they consume up to 50 times the energy per floor space as a typical commercial office building. Globally, data centers are estimated to use between about 90 billion kWh annually, accounting for about 1-3% of the world’s electricity usage.

There has been an increase in the amount of energy used by data centers due to technological advancements and a greater reliance on the internet to carry out work and daily activities. Data centers need electricity to run the equipment and often allocate a significant amount of energy to power servers and keep the devices cool. Although more efficient hardware and innovations in infrastructure and data center cooling systems have been able to offset the growing demand for electricity in data centers, it is believed that the demand for electricity is so strong that it cannot be offset solely by hardware, software, or infrastructure efficiency gains.

There are two types of energy that can be consumed: renewable energy and nonrenewable energy. Nonrenewable energy is power obtained from burning fossil fuels and has a limited supply. Renewable energy includes solar, hydro, and wind energy and is obtained from resources that can replenish themselves at the rate it is used. Energy consumption is calculated the same way for both types of energy, but data centers are shifting to use more renewable energy resources as they are often less expensive and safer for the environment.

How to Measure Data Center Energy Consumption

Data center managers are increasingly focusing on the energy consumption within the facility as the cost of energy is drastically increasing and companies are placing a higher value on green initiatives as they relate to corporate social responsibility. However, it is very difficult to manage something that you cannot measure, so there are numerous metrics beyond a utility meter reading used to measure data center energy efficiency and consumption. These metrics can be used to compare one facility to another and include:

• Power Usage Effectiveness (PUE). PUE is the ratio of the total energy used by a data center and the energy consumed by the IT equipment. Lighting, cooling equipment, and inefficiencies in electricity distribution within the facility are all included in the total energy calculation. Energy consumed by IT equipment is calculated by summing up the power readings of the rack power distribution equipment. Most data centers target a PUE of less than or equal to 1.5, or 1.4 for new data centers.
  
  PUE = Total Facility Power / IT Equipment Power
   
• Data Center Infrastructure Efficiency (DCIE). DCIE is the reciprocal of PUE and is used to convert the PUE metric into a percentage. It is calculated by dividing the total energy consumption of all IT equipment by the entire energy consumption of the facility. This metric is used to identify how efficient the data center IT equipment is in relation to overall data center energy use.
  
  DCIE = IT Equipment Power / Total Facility Power
   
• Corporate Average Date Center Efficiency (CADE).  CADE is a metric used to evaluate and rate the overall energy efficiency of a facility based on its performance. It takes into account the energy efficiency of facilities, their utilization rates, and level of server utilization. Facility efficiency, which is the energy delivered to IT equipment divided by the energy drawn from utilities, is multiplied by IT asset efficiency, which is the average central processing unit (CPU) utilization across all servers until efficiency efforts are undertaken.
  
  CADE = (Facility Efficiency) x (IT Asset Efficiency)

The following table can be used to determine the efficiency of data center energy consumption based on the PUE and DCIE:

There are several locations within a facility where energy consumption can be measured, including the power feed from the utility, the uninterruptible power supply (UPS), floor PDUs, busways, and rack PDUs. Once these measurements are taken, managers can determine if the current KPIs are suitable for the facility. If not, various energy efficiency initiatives may be implemented to ensure a greener data centerv.

How to Reduce Data Center Energy Consumption

Data center managers must ensure efficient operations to not only minimize the harmful effects imposed on the environment but also decrease costs. The more energy used by a data center, the more greenhouse gas (GHG) emissions are exposed to the environment, increasing the data center’s carbon footprint. Also, as data centers expand to manage larger quantities of data, there will be more equipment that needs to be purchased, maintained, and cooled, all of which will increase the amount of money spent on energy.

Some ways data center managers can reduce data center energy consumption include:

• Switch to ECO mode. There is often energy that is technically “consumed” by data centers but is not actually used by any equipment in the facility, otherwise known as wasted energy. UPSs can waste a significant amount of energy unless they are put into Economy (ECO) mode which involves incoming power passing directly through the UPS. In ECO mode, servers run on utility power but are still protected in the event of a power outage. The efficiency of the bypass path when a UPS system is in ECO mode is 98%-99% compared to the base UPS efficiency of 94%-97%. In other words, there is a potential 2-3% reduction in data center energy consumption when the UPS is system operated in ECO mode.
• Use server virtualization. Server virtualization involves physical servers being used as pools of logical computing capacity. Typically, applications are inefficiently deployed across multiple systems where there is a dedicated server and storage for each application. Each of these platforms consumes nearly all the power it would require at peak load, but the platform is doing very little work. Virtualization aggregates servers and storage onto a shared platform while strictly separating systems, applications, data, and users. It dramatically improves hardware utilization and enables the reduction of power-consuming servers and storage devices and their associated cooling equipment.
• Adopt innovative cooling practices. Since cooling systems account for up to 60% of a facility’s utility bill, data center managers must adopt the best practices for cooling. Some of the most efficient cooling practices involve liquid cooling, which leverages the higher thermal transfer properties of fluids to support the cooling of high-density racks. This type of cooling is regarded as being one of the top innovations in data center cooling as it distributes heat over more convection surface area. Other innovative and efficient cooling practices include implementing hot/cold aisle containment sealing cable outputs to minimize bypass airflow and using blanking panels inside equipment enclosures.
• Increase temperature set points. Modern IT equipment can run in warmer environments, but many data center managers still overcool their data centers. Since raising temperatures can result in 4-5% in energy savings for every 1°F increase in server inlet temperature, there is often a large opportunity for increased efficiency by doing so. Implementing environment sensors and environment monitoring software is critical to know how much you can raise temperatures without introducing a risk of equipment damage or downtime.
• Upgrade to new equipment. Although costly, data center managers should consider updating their equipment as some legacy equipment is very inefficient. Newer equipment is often designed to need less power to complete the same amount of work, if not more. Purchasing new equipment can lead to an overall reduction in costs as it may not only lower the energy bill but will also decrease the amount of money spent on maintenance and upkeep typically spent on older equipment prone to breaking down.
• Turn off idle IT equipment. If IT equipment is constantly powered-on, when they become idle, the equipment is still consuming a significant portion of the power it would draw at maximum utilization. To combat this expensive and inefficient occurrence, remotely turn off rack PDU outlets and power down idle equipment to reduce the amount of wasted energy within the facility. Similarly, ghost servers are perpetually idling and should be identified and either decommissioned or utilized.
• Turn on the CPU’s power management feature. The CPU uses more than 50 percent of the power required to run a server. To reduce this consumption, the power-management feature available for most CPUs optimizes power consumption by dynamically switching among multiple performance states based on CPU utilization. This process is done without resetting the CPU and can lead to significant annual savings on energy.
• Use DCIM software. Data center professionals often use Data Center Infrastructure Management (DCIM) software to monitor and manage energy consumption, resulting in more efficient operations. DCIM provides the information needed to implement and enhance many of the measures listed above. With a sufficient DCIM software system, data center managers can have energy data to make more intelligent decisions, get real-time charts and reports on metrics like PUE, create bill back reports to facilitate more energy-efficient behaviors, save energy by avoiding overcooling, identify power-hungry equipment and ghost servers, and intelligently consolidate and virtualize resources. 

Want to see how Sunbird’s world-leading DCIM solution can help you reduce your data center energy consumption? Get your free test drive now!