The European Commission and its consultants have presented a second draft of data center labeling and rating scheme (see Figure 1), as required under the delegated regulation authorization in the Energy Efficiency Directive (EED). However, the proposed label strays from the central objectives of the EED — namely, to report data center information and key performance indicators (KPIs), improve data center infrastructure energy efficiency and reduce the greenhouse gas emissions (GHG) associated with energy use.

Instead, the label presents a laundry list of sustainability metrics. These include energy consumption; combined Scope 1 and 2 GHG emissions; the renewable energy factor (REF) value; PUE and water usage effectiveness (WUE) values; three IT efficiency indicators; heat reuse capability; the energy reuse factor (ERF) value; presence of an end-of-life equipment program; refrigerant type; and the provision of grid flexibility services.

Many of these proposed label reporting items are immaterial to assessing a facility's operational efficiency and offer little actionable or meaningful data to evaluate or modify its performance. Uptime Intelligence has provided detailed comments to the Commission, outlining specific concerns with the labeling proposal and offering alternative approaches and metrics to make it more useful.

Figure 1: Proposed EED data center rating scheme/label

The breadth of the proposed reporting requirements risks exacerbating industry concerns about confidentiality, particularly regarding the disclosure of information and KPIs such as data center local administrative unit (LAU), total energy consumption and idle power metrics. The Commission needs to address these confidentiality concerns under the second delegated regulation and modify the rating scheme proposal to conform with the final confidentiality requirements.

The rating scheme/label (see Figure 1) is organized into three sections (columns), each of which is discussed below.

Origin of energy and resources

Uptime Intelligence has recommended streamlining this section by removing and consolidating some of the information and adding additional information that better addresses the energy and GHG emissions performance of the data center.

Energy consumption and location information

Many operators consider a data center’s energy consumption to be confidential information, making its inclusion as a reported value on the label a point of serious contention between the industry and regulators. Instead, Uptime Intelligence recommended that the label report the data center power capacity — specifically, the number of megawatts of power capacity through the grid connection or on-site power plant, preferably as a range of megawatts (e.g., <1 MW, 1 to 5 MW, 5 MW to 10 MW, etc.). The updated label lists the rated IT power — in kilowatts (KW) or megawatts (MW) — which could be reported as a range and used as the proxy for energy consumption. Providing this information gives a general indication of energy consumption and data center size without providing specific values.

GHG emissions and generation sources

The reported GHG emissions data should be limited to the location-based and market-based REF values, along with the Scope 1 and Scope 2 location-based carbon footprint of the data center’s energy consumption. The REF values indicate the percentage of carbon-free energy consumption in a data center’s energy supply. The location-based value, which reflects the percentage of renewable energy consumed by the data center, would need to be defined, as it is not currently a standardized metric. The market-based REF provides the percentage of consumed and offset carbon-free energy use.

The Commission should designate carbon-free nuclear generation as an acceptable REF source as it is illogical to offset that consumption with guarantees of origin (GoOs) to reach a 100% REF value. These data values sufficiently characterize the carbon emissions profile of the data center’s operation.

Reporting the Scope 1 and Scope 2 emissions values captures the carbon footprint associated with the data center energy use. It is important to report the two values separately because more operators are contracting their standby generation capacity to grid authorities to provide primary power during periods of grid stress. For example, EIRGrid requires operators to provide primary power on request as part of the grid authority’s demand management program.

The generation sources and the market-based electricity emissions factor data offer limited meaningful information beyond what is already provided by the REF and carbon footprint values. In most cases, grid power is the sole generation source consumed by a data center, and the electricity emissions factors are not needed if the overall carbon footprint of the data center is reported. This information can be removed from the label without affecting its utility for assessing data center energy performance.

In the updated label, the Commission has proposed that operators report their market-based energy emissions factor. However, there are two problems with reporting this value. First, 80% of the reporting data centers cited in the third consultation workshop already have a 0 market-based emissions factor and with the Commission’s proposal that all data centers have a 100% market-based REF by 2030, all data centers will report a 0 value. Second, it misrepresents the actual emissions factor of the energy consumed to operate the data center. The Commission should require the reporting of the location-based emissions factor for the data center.

Impact on regional water situation

In the fourth consultation presentation, the consultant modified the water reporting section to display information on the water availability status of the data center — specifically, whether it is located in a water-rich, water-stressed or water-scarce region. The WRI Aqueduct Water Risk Atlas was referenced as the information source. However, the requirement to report the 2050 risk seems unnecessary, as it is outside the operating horizon of a data center facility and not applicable to the reported operating year.

This section would benefit from including the type of cooling system installed (e.g., evaporative, adiabatic, dry-cooler, absorption chiller, hybrid). Identifying the cooling system type will provide a sense of the scale of the water use and enable the Commission to assess WUE thresholds for different cooling system types.

Efficient use of energy resources

The “Energy Efficiency” heading should be changed to “Power Usage Effectiveness.” No other changes were recommended for this section.

Data center performance

The current version of the label requires data center owners to report the total work capacities for standard CPU-based and accelerator/GPU-based servers and the total storage capacity of dedicated storage equipment. The work capacity values are a good starting point for the ICT indicators, but they only assess the available work capacity. They do not address the utilized work capacity, which is the true indicator of operational efficiency.

The Commission should consider adding fields for voluntary reporting of the average utilization of each of the three capacity values. IT operators should be addressing the utilization of their IT equipment, aiming to optimize the capacity utilization to minimize energy consumption while ensuring that the equipment can meet peak workload demands.

The Commission should also consider Uptime Intelligence’s recommendation that the label list two actionable efficiency indicators:

1. The percentage of installed servers in the data center operating at less than 10% average utilization.
2. The percentage of the servers in the data center with the C6-state power management function enabled.

The metrics identify the percentage of poorly utilized servers and the percentage of servers with the most aggressive C-state power management function deployed. Minimizing the first indicator and maximizing the second will optimize the work delivered per energy consumed by the servers.

The storage capacity utilization value provides an indication of the efficiency of the storage products, as their energy use is largely fixed (independent of utilization). The more data stored, the better the efficiency of the storage. These three metrics should be reported for both the current and previous year to assess year-to-year performance improvements.

Colocation operators should not be required to collect or report the ICT efficiency indicators. Instead, the Commission needs to make the IT operators that own this data responsible for reporting the ICT efficiency indicators separately.

Other sustainability actions and data center circularity measures

The reporting of heat reuse availability and ERF values should be retained on the label to identify those operators that have captured this energy reuse opportunity.

Uptime Intelligence recommended that all the “Further sustainability measures” section, which has a list of sustainability measures that operators can voluntarily report, be removed from the label. These items are not appropriate for a label or indicative of data center efficiency. The Commission chose to retain these items on the updated label indicating that they are likely to remain on the final label proposal. The Commission did provide definitions for each of the sustainability items, clarifying the information that operators should report on the label.

The following Uptime Institute experts were consulted for this report:
Dr. Tomas Rahkonen, Research Director Sustainability, Europe, Uptime Institute

Other related reports published by Uptime Institute include:
EED delegated regulation due in May: what’s the rush
Remaining EED reporting deficiencies need immediate attention