Power usage effectiveness or PUE is a standard efficiency metric for power consumption in data centers. A simple definition of PUE is the ratio of total facility energy to IT equipment energy used in a data center and can be represented by the formula:

Total facility energy encompasses the power dedicated to the data center facility or data room, measured at the meter. This includes all loads, including IT equipment, cooling systems, lighting systems, and power delivery components.
Total IT equipment encompasses all energy fed to compute, storage, and networking equipment, including other control equipment like KVM switches, workstations, monitors, and laptops.

Despite the simplicity of the ratio and its acceptance as a standard performance metric, calculating PUE is not as straightforward as the formula seems. There are several practical considerations

Understanding PUE in Data Centers
Generally, PUE is not a one-time measurement. Tracking PUE over time for a single data center will reveal the facility's performance against its initial baseline calculation. Although Uptime Institute reports the average PUE in data centers for 2020 as 1.58, this metric may not be entirely useful for comparison across multiple data centers, each with unique configurations.

Comparing PUE measurements against various data centers, even very similar facilities, proves to be challenging. Two data centers, roughly the same size but in different locations (perhaps different regions or countries) could easily consume power in dramatically different ways for several reasons.
For instance, differences in local climate and weather, power grid services, and even facility building materials may impact energy usage.
Additionally, what is considered by different data center teams as relevant or material in their PUE calculations is not always the same. Teams need to consider how to classify subsystems as IT loads, infrastructure loads, or irrelevant, and even consider if a measurement is practically possible.
Potential challenges for comparing PUE across facilities include:

• Subsystems present in one data center may not be in another
• Subsystems support mixed-use facilities, such as non-data center functions like cooling towers or chillers, and cannot be easily or directly measured
• Instrumenting specific subsystems is impractical and costly, like power distribution units (PDUs), which contain many outlets multiplied by the number of PDUs in use in the facility
• Practical measurement points, such as the meter, may include loads unrelated to the data center and prove to be difficult to separate

How to Calculate Accurate PUE

Where energy measurements are taken impacts how to attribute power usage to its appropriate category, either facility usage or IT usage. For facility readings, take measurements at or near the facility's utility meter. For mixed-use facilities or data centers, target only the meter that powers the data center

Data centers that share a utility meter with non-data center portions pose problems in attributing accurate measurements. In these cases, estimating the non-data center portion and deducting it from total usage may be the only option.

Measuring IT equipment load should be measured after all power conversions, switching, and condition have been applied. Within server rooms, PDUs are the most accurate site for measuring the total power delivered to the server racks.

What Is Data Center Infrastructure Efficiency (DCiE)?

The reciprocal of PUE would measure the energy efficiency of a data center. Data Center Infrastructure Efficiency (DCiE) is that ratio and can be calculated using either of these two formulas.

Those determining DCiE usefulness must grapple with the same challenges of those measuring PUE, markedly that PUE and DCiE are not easily comparable against various facilities, even across similar data centers.

Where the measurements are valuable is when used for benchmarking a data center's power efficiency. If this process is automated through energy performance software and in real-time, both PUE and DCiE can provide a level of insight with significant business applications. By understanding power consumption at various granular levels of the data center, power leaks, and under-utilized systems can easily be discovered, and their costs curtailed.

Other Data Center Efficiency Metrics

With the advent and adoption of metrics like PUE and DCiE, other effectiveness-style metrics from The Green Grid have surfaced as usable measures for data center efficiency.

• Energy reuse effectiveness (ERE) measures a facility's effectiveness in redirecting energy to other uses in the facility, such as using a heat exchanger to convert server discharge heat into heating for surrounding buildings.
• Water usage effectiveness (WUE) attempts to measure the amount of water used by data centers to cool IT assets.
• Carbon usage effectiveness (CUE) is the ratio of the total carbon dioxide (CO2) emissions caused by total data center energy consumption to the energy consumption of IT equipment.