The carbon footprint of cloud-based services is essentially linked to GHG emissions resulting from powering up data centers.
How are GHG emissions computed?
Cloud Computing GHG emissions are obtained by measuring precisely those resulting from the construction of data centers (20%) and the energy consumption needed to power data centers and networks (80%) for any given cloud-based service.
In practice, this is measured by tracking the cloud consumption of a given web service. Major cloud providers provide granular consumption data through APIs, allowing Greenly to measure which vCPU of data centers are used and for how long. These APIs provide minute level information about the instances that are used and in which geography.
Manufacturers of processors like Dell or Lenovo provide detailed life cycle assessments (LCA) of their machines, from which you can compute the electricity output linked to the usage of a vCPU over time.
Greenly's platform converts in real-time vCPU minute level consumption into electricity consumption, expressed in kWh.
Electricity consumption expressed in kWh is then converted into CO2 equivalent, looking at where the data centers are located. This allows Greenly's platform to allocate the correct location-based emission factor (carbon intensity of the electricity grid), expressed in gCO2 / kWh.
Emission factors are typically updated on an hourly basis using an API from the service Electricity Maps.
On top of the footprint linked to electricity consumption, Greenly factors in the footprint resulting from building the data center based on a pro rata of what the consumption of the service represents over the total life cycle of the data center.
Ways to reduce your Cloud Computing GHG emissions
Obtaining such granular analytics allows Greenly's customers to implement engineering solutions for their web services that can reduce their footprint by up to 55%, using the following levers:
Choosing vCPU that emit less (i.e. you don’t need a jet engine for a ride in the park)
Using vCPUs at off-peak hours: at night data centers are underutilized but still run on electricity, thus workload and electricity consumption are lower.
Using electricity at off-peak hours for the same reasons as above
Using greener electricity: selecting a location with more renewable or nuclear power and with less coal and gas in the electricity mix
