Modular Construction & Data Center Design Blog | PCX

How to Predict Your Data Center's Energy Consumption

Written by Jorge Castro | Nov 10, 2020 1:00:00 PM

It doesn’t matter whether you’re building a colocation space or an in-house data center: All facilities must be able to optimize capacity. 

Otherwise, you’ll either end up incurring far more energy costs than you predicted, or you’ll end up with a data center that simply doesn’t have enough power to support your applications, network and data, let alone reliability.

With all this in mind, let’s take a look at some steps you can take to get a better understanding of power consumption and figure out what metrics you need to monitor and track, to avoid running into this potential problem:

1. Understand how power is consumed by your data center, and study the metrics

Predicting data center energy consumption starts with understanding how power is consumed across a data center. This means you need to prioritize learning the terminology that’s relevant to data centers, while studying and optimizing relevant metrics to ensure that you’re operating efficiently.

One study found that IT equipment is responsible for 45 percent of a data center’s energy consumption, while cooling equipment accounts for 39 percent and the rest of the data center is responsible for the other 16 percent. In other words, if you’re trying to reduce your energy consumption, just flipping the lights off more frequently isn’t going to move the needle that much.

You also need to have a working understanding of the various terms and metrics you’ll come across as you try to predict energy consumption in your data center:

  • Power utilization effectiveness (PUE) describes how much energy your IT equipment is responsible for consuming in your data center. To account for seasonal fluctuations (e.g., more power will be needed in the summer versus the winter), this metric should be measured over a 12-month period, with a target goal of 1.2.
  • Coefficient of performance (COP) refers to the amount of heat dissipated by a data center, divided by the amount of energy needed to move it. Ideally, this number should be between 3 and 4. The closer this number is to 1, the closer your energy costs to power your servers. Your energy costs to cool them will be the same.
  • Return temperature index (RTI) is the ratio of airflow needed to cool servers and IT equipment, divided by the total amount of air the cooling system supplies. In a perfect world, this number would be 1.

Although the above list isn’t exhaustive, you should be able to get a pretty good handle over your data center energy costs by studying these metrics, while working with your team and partners to improve them over time.

2. Locate and remove unnecessary equipment running in the background

If you’re running legacy servers and other equipment that your team is no longer relying on, you are burning up way more power than needed. 

To lower your energy consumption costs, take a quick inventory of your data center to make sure that all unnecessary equipment is taken offline. In addition to lowering your electric bill, you’ll also be lowering your carbon footprint.

3. Ensure you’re using all energy as efficiently as possible

In recent years, the number of data centers and the workloads they manage has risen considerably. Over that same period of time, however, the amount of energy that data centers consume hasn’t increased all that much

In large part, this is due to the fact that data centers are being built with energy-efficient technology and that more and more organizations are leveraging cloud computing and hosting their workloads in state-of-the-art data centers controlled by hyperscale providers.

If you’re looking to build an energy-efficient data center, you should consider the modular approach to construction. By working with the right provider, you will be able to accelerate construction timelines and reduce budgets, while lowering your PUE and ensuring you’re using the most energy-efficient components for your project.

Using prefabricated components makes predicting data center energy consumption easy. By partnering with PCX, you’ll end up with an energy-efficient data center that’s engineered to the highest quality standards and arrives on your property fully tested and ready to go.

To learn more about why more and more companies are choosing to take the modular approach to data center construction, check out our free guide: The Complete Guide to Modular Data Centers.