Operational emissions are emissions caused by energy consumption (Green Software Foundation, 2024). In cloud computing, this means the energy consumed to power the hardware components hosting a server instance, and other associated data center equipment.
Tailpipe’s approach to measuring the operational emissions of cloud computing is underpinned by the Software Carbon Intensity (SCI) Specification. The SCI Specification is internationally recognized as the ISO 21031 specification for calculating software’s carbon emissions. According to the SCI Specification, to measure operational emissions:
‘multiply the electricity consumption of the hardware the software is running on by the region-specific carbon intensity.’
To calculate the operational emissions of cloud computing, Tailpipe multiplies the energy required to run the organization’s cloud computing needs by the total carbon intensity of the electricity grid powering the data center where the organization’s cloud services are hosted.
Operational emissions are measured in kilograms of carbon dioxide equivalent (kgCO2e). CO2e is a unified measure of global warming potential (GWP) that accounts for the different greenhouse gases produced during the energy generation process that powers your cloud computing usage.
Tailpipe Overall Operational Emissions Methodology
Tailpipe calculates the kilograms of CO2e produced from the organization’s use of cloud computing using the following formula:
((((Computing Infrastructure + Network Storage + Intra-Region Data Transfer + Intra-Region Non-compute Networking) * PUE) + (Inter-Region Data Transfer + External Data Transfer + Inter-Region Non-compute Networking + External Non-compute Networking)) * Carbon Intensity * Power Transmission Losses)
Where:
Computing Infrastructure = (Processors + Accelerators + Memory + Onboard Storage + Motherboard) * Instance Power Supply Efficiency Factor
Network Storage (Wh) = ((Power Draw of Provider Network Storage * GB of Network Storage) * Hours of Utilization) * Network Storage Power Supply Efficiency Factor
Customer Data Sources
Detailed Usage and Billing Information
To calculate accurate operational emissions figures, Tailpipe starts by analyzing the organization’s detailed usage and billing information from the Cloud Service Provider (CSP). For AWS, this data is gathered from the Cost and Usage Report (CUR) files. For Azure, this is the Command Line Interface (CLI) and account invoices. These files are generated by the CSP to show their customers which cloud services they have used and what they will be billed for.
These files also provide the start and end time of the customer’s usage of each virtual machine. Tailpipe uses these time periods to calculate the specific operational emissions associated with usage.
Virtual Machine Utilization Data
The quantity of operational emissions produced by an organization’s cloud computing infrastructure is not only measured by the length of time that the instance is used for. Another important factor is utilization: the percentage of the infrastructure’s capability that is required to carry out tasks. For example, processing a large dataset would use a high amount of a virtual machine’s processing power, whilst displaying a static image uses very little.
AWS provides utilization data through its AWS CloudWatch service. By default, CPU utilization is provided for cloud instances and is refreshed every minute (N.B. this can be changed to refresh every 10 seconds, every two minutes, every five minutes or every 15 minutes).
Azure provides utilization data through Azure Monitor, which is also refreshed every minute.
Tailpipe takes the virtual machine utilization data and averages it over the same period as the appropriate line item in the usage and billing report which, in AWS and Azure’s cases, is an hour.
Data Quality
Tailpipe draws on a range of internal and external sources to calculate operational emissions and only uses sources it has high confidence in. This is shown in Tailpipe’s Data Quality Dashboard.
Internally, Tailpipe has lab tested a range of server configurations, to obtain valid data on the power consumption of their components.
Externally, Tailpipe uses carbon intensity grid mix data from Ember or national sources where available. It sources the rest of its external data from component manufacturer datasheets and specifications, CSP reports, and journal-published computer science studies. Every data source that Tailpipe leverages has been detailed throughout this documentation, and in its Appendices.
For a detailed breakdown of Tailpipe’s data sources, view the Data Quality Dashboard.
