Questions about Boavizta Carbon Footprint Calculation Methodology

[joanacmferreira]

Hello,

Hope you are well.

As suggested by David Ekchajzer, I am posting my questions related to Boavizta Carbon Footprint Calculation Methodology here:

CPU power consumption profile:

1. Is Boavizta CPU workload definition the same as Ganglia metrics derived CPU workload? (number of processes that are running on the system plus the number of processes that are waiting to run normalized to the number of cores and capped to 100 %)
2. Is it possible to derive CPU workload from CPU utilization?

RAM power consumption profile:

For Ram Boavizta has a value that represents the power consumption for that instance independently of the RAM workload. But when I investigate the RAM power consumption profile in Teads dataset I see a linear relation between RAM workload and Power consumption. This means that Boavizta most recent findings indicate that this relationship no longer exists?

Other components:

Boavizta defines a 33% consumption for the other components while the Teads dataset has a delta value specific to each instance. 1. Why did you opt for the percentage?
2. Does that percentage account for the GPU

Cloudwatch metrics

Teads mentioned in a medium post that it was able to calculate the Carbon Footprint from CloudWatch metrics. Do you have documentation relative to the methodology?