How to Measure and Reduce Data Center Energy Consumption
Data center energy consumption is on the rise as the demand for computing power and digital services continues to grow.
To reduce data center carbon emissions, many data center managers now have corporate sustainability goals that they must comply with. According to KPMG, 80% of companies now report on their sustainability.
The benefits of increasing data center energy efficiency are vast. A sustainable data center reduces the facility's impact on climate change, reduces operating costs, maximizes the value of existing capacity, complies with regulations and initiatives, and is more attractive to customers.
But many data center managers don't know how to measure their data center energy consumption and intelligently reduce it without introducing risk.
Fortunately, there are some best practices that any data center manager can follow to have a greener data center.
1. Measure and monitor your data center energy consumption
The first step to decreasing data center energy consumption is to measure it. Without key information about your energy usage, you won't know what your baseline is, how much progress you make, or where your opportunities to increase data center energy efficiency are.
Intelligent rack PDUs are a must-have for measuring power usage and available capacity at each rack. Outlet-metered PDUs provide data at the device level which comes with the added benefits of helping you identify ghost servers and power hogs, know energy usage and cost by customer, and compare IT efficiencies. Switched PDUs with outlet control are also useful for remotely powering on and off individual outlets so you can save energy when equipment is not in use.
Other power meter options include bus drops and busway end feeds, remote power panels, floor PDUs, UPSs, and building meters. The data from these items helps improve capacity and reporting on energy efficiency metrics.
Next, you need to collect, store, trend, and report on all this data to extract the most value from it and intelligently reduce your data center energy consumption.
Data Center Infrastructure Management (DCIM) software collects the live measured readings from power meters and transforms them into actionable insights that make reducing your data center carbon footprint easier and faster.
Modern DCIM software can collect billions of data points from your data center every day and distill them into easy-to-understand and pre-configured dashboard charts, reports, and 3D visual analytics to simplify data center energy management.
With intelligent power infrastructure and DCIM software, you can easily measure your data center energy consumption, energy cost, carbon footprint, and much more. By trending this data, you can see the impact of your energy efficiency initiatives in real-time.
2. Measure and monitor your data center environment
The data from environment sensors provides useful information such as where you can improve efficiency and by how much without putting your equipment outside of recommended ranges to maintain uptime.
Environment sensors are cost-effective and easy to install. The sensors that help increase data center energy efficiency include:
- Temperature. Install temperature sensors in each rack to know if you are overcooling and by how much. Then, you will know how much you can raise temperatures to save energy.
- Humidity. Deploy one humidity sensor for every five racks to know if you are excessively humidifying or dehumidifying the data center.
- Airflow. Place an airflow sensor at each cold air supply and hot air return to ensure the cooling system is functioning properly and efficiently.
- Air pressure. Differences in air pressure between hot and cold aisles could lead to partition leaks that reduce efficiency. Air pressure data from under the raised floor can feed into cooling systems or Building Management Systems (BMS) to vary fan speeds to increase efficiency.
The same DCIM software that transforms your power and energy data into actionable information in step one should be used to collect and report on environment sensor data. For example, you can view thermal or humidity time-lapse videos on your floor map in 3D to easily spot wasteful overcooling or leverage out-of-the-box KPI charts like delta-T per cabinet.
3. Track PUE and other data center sustainability KPIs
If you've completed steps one and two, you now have a fully instrumented and monitored data center. You can now begin tracking the data center sustainability KPIs that indicate how efficient your data center is and where it's going.
Power Usage Effectiveness (PUE) is one of the most popular data center KPIs. It is the ratio of total facility energy to total IT equipment energy. The closer to 1.0 your PUE is, the more efficient your data center is. If you have a newer data center with state-of-the-art technology, your PUE target should be around 1.2 but no more than 1.4. If you are optimizing an older facility, aim for a PUE of 1.5 or less.
Calculating PUE can be a tedious process full of manual meter readings and math, but with DCIM software, you can automatically calculate, trend, and compare the PUE for all your sites in real-time. Then, you can focus on strategic projects that will reduce data center energy consumption rather than time-consuming and potentially inaccurate PUE math.
There are many metrics that measure different aspects of data center energy consumption and sustainability. For a holistic picture of how green your data center is, measure the most relevant ones to you such as:
- Percentage of Cabinets Compliant with ASHRAE Standards – Measures how much progress has been made towards ideal environmental conditions.
- Carbon Usage Effectiveness (CUE) – Measures the overall sustainability from a data center carbon footprint
- Cooling Capacity Factor (CCF) – Measures the overall efficiency of your data center's cooling system.
- Data Center Space Efficiency (DCSE) – Measures how efficiently your data center space is used.
- Delta-T Per Cabinet – Measures how effectively cold air is cooling the IT equipment in each cabinet.
- Green Energy Coefficient (GEC) – Measures how much renewable energy is generated onsite.
- Stranded Power Capacity Per Rack – Measures how much additional equipment can be deployed in existing cabinet resources.
- Technology Carbon Efficiency (TCE) – Measures how clean the energy consumed by your data center is.
- Water Usage Effectiveness (WUE) – Measures how efficiently water is being used in the data center.
4. Avoid overcooling your data center
ASHRAE's thermal guidelines state that temperatures should be between 65-80°F at the server intakes. However, many data center managers are understandably risk-averse and keep their equipment much cooler.
Since you can reduce data center energy consumption by 4-5% for every 1°F increase in server inlet temperature, there are massive potential energy savings for most data centers.
With temperature sensors and DCIM software, it's easy to safely increase temperatures while ensuring that the equipment is safely within recommended guidelines.
To avoid overcooling and wasting energy, follow this simple process:
- Per ASHRAE guidelines, deploy temperature sensors at the top, middle, and bottom of both the front and back of each rack.
- In your DCIM software, set thresholds for rack inlet temperatures so you are automatically notified when temperatures exceed your chosen temperature.
- With your DCIM tool's built-in psychrometric cooling charts, determine if your racks are within your chosen guidelines. If not, assess how much you can raise temperatures.
- Slowly raise your temperature set points while checking your cooling charts to ensure you don't exceed your thresholds. In some cases, you can even turn CRAH/CRACs off entirely.
- Occasionally review your temperature trends and thermal time-lapse videos in your DCIM and make any necessary adjustments.
5. Implement data center containment
Separating the cooling units' cold supply air from the hot exhaust air from IT equipment can reduce data center energy consumption by 30% or more. This is because when hot and cold air mix, the cooling infrastructure must waste energy by cooling air that is already cold.
Consider a data center containment and monitoring strategy to keep hot and cold air from mixing. This will enable a uniform supply air temperature and a warmer return air so that your CRAH/CRACs will only cool the warmest air in the data center.
There are several different ways to achieve this. Popular containment options include:
- Hot aisle containment. Ceiling panels and doors enclose the hot aisle between rows of cabinets so that warm exhaust air can be separated and returned to the cooling system.
- Cold aisle containment. Ceiling panels and doors enclose the cold aisle between two rows of cabinets, allowing the rest of the data center to become a hot-air return plenum.
- Chimney systems. The warm exhaust air is pushed up through chimneys above each rack to the ceiling return air plenum.
- Curtain systems. Racks are faced in alternating rows of front-to-front and back-to-back and curtains separate the cold supply air and warm exhaust air.
6. Keep bypass airflow to a minimum
Airflow management is an often-overlooked way to decrease data center energy consumption. By optimizing airflow, you can further boost the efficiency of your cooling infrastructure.
The goal is to minimize bypass airflow which is air that does not cool IT equipment before returning to a cooling unit. Bypass airflow accounts for about 60% of all airflow on average, so there is ample opportunity to increase efficiency, and it is often fairly inexpensive and fast to achieve.
Some ways to reduce bypass airflow in your data center include:
- Blanking panels. Blanking panels are a low-cost solution that improves airflow, decreases server inlet temperatures, and increases return air temperatures. Panels alone can reduce energy costs by 1-2%. Don't forget to plug up holes such as open U spaces, spaces between rails and edges of cabinets, and under cabinets.
- Raised floor. Perforated tiles should be placed on the raised floor in front of IT equipment that requires cooling. Use grommets in the raised floor to seal cable penetrations and allow for maximum pressure to be maintained under the floor.
- High-flow cabinet doors. Perforated cabinet doors improve airflow compared to doors made of glass or similar materials.
- Efficiency hoods. Efficiency hoods for your CRAC units allow them to receive warmer air by extending your unit up to the return plenum. Pulling the hottest air possible out of the data center causes the temperature to get cooler and to cool quicker. Hoods can increase the capacity of cooling units by up to 25% and allow CRACs to cycle off more frequently, extending their lifespan.
7. Find your ghost servers and power hogs
Ghost servers are idle servers that perform no useful function while still consuming energy and space. It's not uncommon for up to 30% of all servers in a data center to be ghost servers.
With the right tools, identifying ghost servers is as easy as clicking a button.
If you have outlet-metered intelligent rack PDUs and DCIM software, you can run a ghost server report. Simply enter a date range and you'll get the maximum load, average load, minimum load, and total for each device in that time. Then, you can either power off, decommission, or make use of those servers to save energy and maximize the utilization of your existing resources.
Just as easily, you can see which servers are your power hogs that consume much more energy than your other devices. These devices are good candidates to be replaced with more efficient hardware or virtualized.
8. Consolidate your data center footprint
Older data centers or data centers that are underutilizing their existing capacity can benefit greatly from consolidation and virtualization that reduce the number of physical assets you have and the resources they consume.
Data center consolidation reduces energy consumption because there is less equipment and sites to power and cool.
The key to a data center consolidation is to have highly accurate capacity planning that enables you to fit as many devices as safely possible within each of your racks. Many data center managers fail to achieve this because of shortcomings in the traditional approach to device power budgeting.
It's common practice to derate a server's nameplate value to around 60-70% to allow for a safety margin. However, this process is based upon estimations and is largely inaccurate. It results in stranded power capacity and wasted space that negatively impacts your efficiency.
The new way forward is to automate rack power capacity planning to get the most out of your existing resources and defer building out new cabinets or even new data centers.
With outlet-metered PDUs and a modern DCIM solution with an "Auto Power Budget" feature, the software can automatically calculate an accurate power budget number for each make and model instance of a device based upon the actual load of that device in your environment running your applications.
Auto Power Budget provides you with many opportunities to safely deploy more devices in fewer racks, enabling highly efficient data center operations.
DCIM users like Comcast and eBay report improvements in rack power utilization by as high as 40%. For more details, read the Comcast case study.
9. Drive more efficient behavior from your customers
Ensuring that your internal or external customers comply with energy efficiency best practices can be difficult.
One way to encourage them and drive a culture of environmental awareness and accountability is to charge them for the energy their equipment consumes. Then, when they see how their behaviors result in higher costs, they are much more likely to comply with best practices.
DCIM software can collect and display energy and cost data in the form of bill back reports. These can be delivered by customer, data center, or even by rack. When you can report on costs by customer, business unit, or application, it's very easy to identify power hogs. Plus, your charges are always accurate.
Bringing It All Together
As more organizations set goals to become carbon neutral, data center managers must ramp up their efforts to report on and reduce data center energy consumption.
By deploying the right hardware and software tools and following best practices, any data center manager can operate a more sustainable data center.
For more information, check out our new eBook Increasing Data Center Sustainability: 10 Best Practices to Reduce Your Carbon Footprint.
See how Sunbird's world-leading DCIM software can help you measure and reduce energy consumption. Try it free now.