Network Energy and GHG Approximation Methodology

Akamai owns and operates its network IT equipment and outsources data center operations to third-party collocation data centers. Currently detailed, comprehensive, and direct energy consumption data are not available for Akamai’s IT equipment, or the third-party data center infrastructure in which they are hosted. Using existing internal and supplier-provided data the following methodologies were developed to estimate the monthly Akamai network-wide energy consumption.

Server and Networking Equipment Energy Estimate:

  • Power draw (Watts) is measured in the lab at peak load for each server type and configuration (e.g., # of disks). It is assumed that this peak power consumption is the same for a given server type and configuration. This is also done for network switches, routers and PDU’s.
  • At the end of each month an equipment inventory is taken for each data center (including switches, routers and PDU’s) for equipment that is powered on. The number of each server type and configuration is summed for each data center. It is assumed that this equipment has been resident in the data center for the entire month.
  • Each data center’s server type and configuration total is multiplied by the peak power consumption (from #1), multiplied by 24 hours, multiplied by the number of days in the target month, and divided by 1,000 to convert to total KWH for that server type and configuration in a target data center.
  • The total kWh is reduced by a percentage for each server type that reflects the findings from production data of average daily variation of server power draw relative to peak load (from #1). For example, the percentage reduction could range from 15-40%.
  • This per data center server network monthly energy consumption is uploaded into Akamai’s energy and carbon management system.
  • The energy and carbon management system converts the energy usage into GHG emissions by applying both market- and location-based electricity emission factors based on each data center’s location.
  • The Scope 2 GHG emissions for each data center are then summed for total GHG emissions across the Akamai server network.

In 2014 the method used for estimating network IT electricity usage was modified which resulted in a more accurate assessment of total network electricity consumption. Based on extensive production data analysis the average power draw factor was reduced from 85% to 60% of peak power for server generations deployed beginning in May 2010. These server generations now make up the majority of network server deployment. This change was applied retroactively and resulted in a significant reduction in total network-related electricity consumption and Scope 2 and Scope 3 emissions.

Collocation Data Center Energy Estimate:

Collocation data center infrastructure includes but is not limited to backup power supply (UPS), power distribution units (PDU’s), transformers, cooling units, chillers, fans, and lights.

For the estimation of the Scope 3 GHG emission attributable to our third-party data center hosting operations, we rely on the estimated Scope 2 emissions of our server network, as detailed above, and the power usage effectiveness (PUE)i of these data centers to arrive at estimated Scope 3 emissions of these data centers. Where available the PUEii reported by a data center provider for a facility is used. Where not available, the average of reported PUE values is used.

  • For each data center the monthly total server network electricity consumption is multiplied by one minus the data center PUE value:
    Data Center Monthly Electricity Consumption = Server network monthly electricity consumption x ( data center PUE - 1)
  • This per data center monthly electricity consumption is uploaded into Akamai’s energy and carbon management system.
  • The energy and carbon management system converts the energy usage into GHG emissions by applying both market and location-based electricity emission factors based on each data center’s location.
  • The Scope 3 GHG emissions for each data center are then summed for total Scope 3 GHG emissions across the Akamai network.

Categorization of GHG Emissions

Under the Operational Control consolidation approach, Akamai categorizes as Scope 2 the GHG emissions associated with the electricity consumption of our server and network equipment:

  • Akamai is directly involved in the design of our network server equipment, including efficiency.
  • Akamai purchases, owns and has sole control over this network server infrastructure.

We categorize as Scope 3 the GHG emissions associated with the support services provided by our third-party data center hosting providers, including cooling, lighting, power backup and conditioning, and building operations:

  • Akamai has no direct operational control over these operations or their efficiency.
  • These support services are paid for indirectly as part of our data center hosting agreement.

iPUE, as defined by The Green Grid, is the ratio of the total data center energy consumption and the IT equipment energy consumption (i.e., our servers and switches). Ideally PUE measurements are time averaged.
iiReported PUE’s are sourced from an annual survey that Akamai conducts of major collocation data centers representing >75% of Akamai's network server deployment