4. Methodology Overview
How the methodology works at a high level.
Greenpixie's cloud emissions methodology calculates the environmental impact of public cloud workloads as energy, carbon, water, and embodied emissions metrics for each unit of cloud usage.
At the centre of the methodology is a hardware mapping model. Every cloud SKU, regardless of provider or service type, is mapped to the physical hardware components responsible for delivering each unit of work. This mapping spans compute, storage, and networking, covering processor types, accelerators, memory, multiple storage media tiers, and network devices at several layers of the data centre network stack. The energy for a unit of cloud usage is built up from the energy of the hardware components it maps to.
How each component's contribution is determined depends on how much of the underlying hardware is known. Where the hardware is specified, such as a compute instance with a defined processor, vCPU count, and memory, the components are taken directly from the published specification. Where the billing unit is more abstract, such as a managed service whose hardware is not exposed, the hardware profile is determined through research into cloud provider documentation, pricing structures, and public technical disclosures. This research is described further in Section 5.
This approach allows the methodology to cover the full breadth of cloud billing catalogues, from compute instances where the hardware mapping is direct to abstract managed services where the relationship between billing unit and physical hardware is less obvious.
The methodology operates through four parts, being IT electricity usage calculation, facility energy and emissions calculation, embodied emissions calculation, and water consumption calculation. Section 7 describes each in detail. Together they produce the environmental metrics for each unit of cloud usage.