IT Pros: tools for mitigating cloud carbon footprint
Programmer Tools
Green coding tools
Cloud service hosting
Green hosting companies and low-carbon options at AWS, Azure and Google Cloud
Emissions modeling
Green Software Coding
Track emissions and enable emissions tracking capability
Our list of tools from across the web
Awesome Green Software List
Visit the “Awesome Green Software” list for an up-to-date compendium
Green Software Practitioners Training
Linux Foundation’s training course to learn green software methodology
Software Carbon Intensity Guide
Calculate the SCI score for a software application using the Software Carbon Intensity metric
The landscape of cloud emissions tools
These tools are hard to find
The field of cloud carbon footprint calculating tools is still fairly nascent. It contains web-based emissions predictors, Big 3 impact trackers, third party impact trackers, middleware, green coding tools, and online calculators for checking public websites.
As Ross Fairbanks wrote of cloud carbon footprint tracking from the big cloud providers in the Green Web Foundation blog (April 24, 2024), “Right now, most of the metrics shared by most cloud providers make seeing the impacts these changes difficult at best – we really need higher time resolution data to be available to tell if what we’re doing is helping, and standardised measurements for the carbon intensity of compute would really help here.”
The Big 3 carbon trackers are not sufficient
The tools that most cloud customers will end up using are the ones the Big 3 provide in their cloud tools suite. These tools estimate retrospective greenhouse gas emissions based on billing and usage statistics. They do not offer estimates for future emissions of use cases a customer could try out, with variables such as time of use, equipment selected or application choice. So they do not aid customers in planning the best steps for reducing their environmental impact.
The results these trackers do provide would be useful for greenhouse gas accounting by customers, but only if the data and methodologies are shared for peer review and evaluation. At the time of this writing, the methodologies and most of the input data have not been published by the Big 3.
Accurate results for customers depend upon accurate data being used in the calculations, and that can only be verified if shared. For instance, information about the power supply is important information to share. Power is the biggest contributor to cloud GHG emissions, so if the electricity powering the data center that hosts customer’s instance is mostly renewable, the carbon footprint is reduced.
Google Cloud provides customers with the carbon intensity of the electricity powering their cloud instance. AWS and Azure do not. Google is the only one offering the actual emissions of the physical data center, by tracking electricity use by location, while AWS and Azure generalize across the clean energy they bought without specifying location.
Other key data for calculating the carbon footprint of a cloud instance are not shared by any of the Big 3, such as granular power efficiency coefficients or embodied emissions. The Big 3 do know this information and they might use it in the tracking tools they provide their customers, but the inputs to the tools’ calculations are not shared. Azure and Google Cloud provide Scope 3 emissions and AWS does not.
Some researchers are verifying the Big 3 carbon trackers
Benjamin Davy published a Medium article for Teads Engineering about calculating the embodied emissions of a Big 3 instance. He built a repository of hardware and networking equipment, then sourced emissions data from the equipment manufacturers. He used the approach to calculate embodied emissions for AWS, Google Cloud and Azure, making use of the Cloud Carbon Footprint calculator.
Third-party tools such as Cloud Carbon Footprint and Green Algorithms rely on independent data sources and reports to estimate cloud carbon emissions. For power, they use the average carbon intensity of power supply in a geographic region. If one tried to validate the results of a Big 3 carbon tracker using one of these third-party tools, it would be hard to get an apples-to-apples comparison. Big 3 facilities often source low carbon power directly from generation plants or via power purchase agreements. Even if the renewable power they purchase is far away from their data centers, they most likely use a lower carbon figure for their customer cloud impact tracking tools.
But since Google Cloud is more transparent than the other two brands and shares their power mix, one verification test is possible. Cloud Carbon Footprint used GCP power figures to calculate emissions from a use case on both their own tool and on the GCP tool. They analyzed how closely their results matched the Google Cloud tool’s results, and found the numbers were close. They had no similar data to use for comparisons against AWS or Azure.
We believe the Big 3 should share the methodology of their emissions tracker calculations and give visibility into the data used. We also believe that, like all cloud carbon calculators, the Big 3 tools should include predictive estimates based on use cases.
Cloud Service Hosting
Low-carbon cloud platform options
Green hosting companies
But the listing is broad enough to include companies that don’t transparently substantiate their clean energy claims or publish a sustainability policy, as well as some that only use offsets for a 100% clean energy claim.
Make your hosting selection carefully. Skip the companies that only use offsets and focus on those with a direct supply or a demonstrated source of renewable energy. Make note of other policies that impact the environment, such as water reuse and power/operational efficiency.
Low-carbon options at AWS, Azure, and Google Cloud
AWS, Azure, and Google Cloud differ in helping customers identify options for lowering carbon emissions. Google offers noncustomers visibility into which locations are low carbon, while AWS and Azure do not. Their Region Picker tool is useful.
But there are independent sources that evaluate the carbon intensity of the grid power for hyperscaler cloud services. Their information is not current, but the ease of comparison is valuable.
Please note this information is from 2023 and is missing newer locations. But it is useful and valuable, since the top data center location selections below are currently still excellent.
AWS offers the cleanest powered data centers of the Big 3, but says its customers need to do the legwork to find them.
The AWS Sustainability Blog says AWS matches 100% renewables within 22 regions, but no detail is given by Availability Zone. A noncustomer trying to dig in on details will discover the website actually says customers have to do the legwork. Amazon has utility scale renewable projects but give scattered updates about them. In many cases AWS buys certificates for clean power within the same grid as its data centers. They do not specify matching the power within the location and time frame of their actual power use.
Climatiq uses data from Cloud Carbon Footprint and its own emissions factors database to report the carbon intensity (aka clean power) for AWS Availability Zones. AWS has the lowest emissions possible at eu-north-1 in Sweden, effectively none. It is 10 times cleaner than Google’s cleanest location and 32 times cleaner than Azure’s (although those locations are still very good). AWS’ eu-west-3 in France, sa-east-1 in Brazil, ca-central-1 in Canada, eu-est-2 in England, and eu-south-1 in Italy are also very good.
Be cautious with AWS Asia and Middle East options because their power is high emissions. The very popular us-east-1 is located in Virgina, where power of any kind is in increasingly short supply, and its middling grid carbon intensity is looking to get worse as new fossil powered plants come online to meet overwhelming data center power demands.
Azure is the laggard in low carbon emissions locations and requires its customers to do complicated research.
Azure is targeting 100% of its data centers to be powered by renewable energy in 2025. They do not offer detail about the carbon intensity of availability zones to the public. Customers of their “Pro BI” can look at and make selections based on low-carbon criteria.
The Climatiq analysis shows Azure does not have a location as low as AWS’s best 6. Azure’s best, uk-west and uk-south, are 32 times more carbon intensive than AWS’s eu-north-1 in Sweden. The next best from Azure is north europe in Ireland. Those are the top ones worth mentioning.
The worst Azure locations for carbon intensive grid power are in India.
Google Cloud leads in sharing information and in progress toward 24-7 matching, with good low carbon options.
Google Cloud’s public website identifies their low carbon locations, so noncustomers are able to get this important information. Google’s goal is to match their actual electricity use with an identified supply of low carbon electricity for each location. Their transparency exceeds the other Big 3 brands.
Google Cloud’s location with the lowest carbon intensity grid power is erope-west6 in Switzerland. It is very low but still 10 times higher than AWS’s eu-north1 in Sweden. Its next best location is southamerica-east1 in Brazil, and its uswest-1 in Oregon and its northamerica-north in Canada are both good choices.
Google also identifies unattended projects in its Active Assist tool to suggest possible reductions in cloud usage.
Green Algorithms
Find the lowest carbon configuration for a project
Predictive calculator
Input a use case and get emissions. Cloud providers and availability zones/locations, run time, equipment. Thanks to the predictive calculator published by the Green Algorithms Project, there is a way to identify candidates for where to mount your instance.
The Green Algorithms project was jointly developed by Loïc Lannelongue, Jason Grealey, and Michael Inouye
This team points out:
“The main factor impacting your footprint is the location of your servers: the same algorithm will emit 74 times more CO2e if run in Australia compared to Switzerland.”
Green coding tools from the web
1. Emissions use case forecasters and emissions trackers
Name
Type
Description
URLs
Green Algorithms
Computing use case emissions forecastor
Green Algorithms is an open source project dedicated to promoting sustainable computing. Their online calculator estimates the carbon emissions of a computing job with a specified location, compute hardware, storage and runtime. The information helps users decide the best way to reduce their impact before setting up an instance. The results provide carbon footprint and energy consumption along with the EPA’s GHG equivalencies calculator’s descriptive impacts such as kilometers in a passenger car. There is also a community of practice.
GA4HPC
HPC use case emissions forecastor
GA4HPC is from Green Algorithms, and it is a script to pull the logs from any High Performance Computing Cluster, and allows users to enter a use case to get the carbon footprint estimate. It isn’t a web-based calculator such as the Green Algorithms, but it is a useful tool for anyone who can insert it in SLRUM as a workload manager.
ML CO2 Impact
GPU emissions estimator
ML CO2 Impact is a machine learning cloud impact calculator that estimates GPU carbon emissions for AI applications. The calculator collects hardware, compute time and cloud platform region, and utilizes CodeCarbon‘s middleware. The kg CO2 results also include EPA’s GHG equivalencies calculator’s descriptive impacts such as kilometers in a passenger car. The authors want it to be used in particular with research studies to publicize the impact of training models.
Cloud Carbon Footprint
Computing use case emissions forecastor
Cloud Carbon Footprint (CCF) is an open source tool to estimate energy use from public cloud usage, providing greenhouse gas emissions and energy consumption for a user-defined cloud computing use case. CCF provides public repositories of data on the carbon efficiency factors for Amazon Web Services, Google Cloud Platform and Microsoft Azure as well as for specific brands and models of hardware. Their data sets are made public via the API provided by Climatiq (below).
Climatiq
Software emissions estimator API & data
Climatiq is a Berlin-based certified B Corporation that sells emissions data sets and supporting services enabling their customers to add GHG emission calculations and auditing to software programs. Climatiq also generously publish an API for the open source emissions factors data set from CCF (above). They have blog posts that explain a great deal about how to measure the environmental impact of operating data centers or using the cloud.
AWS Customer Carbon Footprint
AWS Customer Carbon Footprint does not disclose methodologies or key data, including power carbon intensity. It provides results to its customers from billed usage.
AWS does not provide noncustomers with information about low carbon availability zones.
Google Cloud Carbon Footprint
Google Cloud Platform Customer Carbon Footprint is available to GCP customers. It helps calculate the carbon emissions of workloads that have already been run and billed. It provides results to its customers from billed usage but does not disclose full methodology or data, although it does share availability zone grid carbon intensity.
Uniquely among the Big 3 Google Cloud Platform allows noncustomers to explore which locations are low carbon. We applaud this transparency and believe it will encourage sustainability.
Microsoft Azure Emissions Impact Dashboard
Big 3 customer tool
Microsoft Azure Emissions Impact Dashboard is available to Power BI Pro customers. It helps calculate the carbon emissions of workloads that have already been run and billed. It does not disclose the methodologies or key data, including power carbon intensity. It provides results to its customers from billed usage.
Azure does not provide noncustomers with information about low carbon locations.
2. Green coding tools
This list identifies green software development tools that add emissions measurement to a software application or test the emissions from running an application.
Name
Type
Description
URLs
Carbon Aware
Software emissions tracking and use cases
Carbon Aware is an open source testing framework from the Green Software Foundation to measure the carbon emissions of software and to run use cases on where/when to run it.
API emissions tracking coding tool
CO2.js is an open source Javascript library from the Green Software Foundation to test the carbon footprint of APIs. It could be used to check and block the uploading of carbon intensive files. It could also present users with information about the carbon impact of their online activities.
CarbonTracker
AI power consumption and emissions modeling software
CarbonTracker enables AI applications to track and predict energy consumption and carbon footprint of a GPU while training deep learning models. It is an open source tool from the Green Software Foundation.
CodeCarbon
AI emissions tracking software
CodeCarbon enables tracking the carbon intensity of AI code bases. It is a lightweight software package that integrates into Python codebase. It estimates the amount of carbon dioxide produced by the cloud or personal computing resources used to execute the code. It also shows how to lessen emissions by optimizing code or hosting cloud infrastructure in geographical regions that use renewable energy sources.
Experiment Impact Tracker
AI power consumption and emissions modeling software
Experiment Impact Tracker is a drop-in that enables AI applications for NVIDIA GPUs and Intel chips to track power draw, hardware information, python package versions, estimated carbon emissions information, etc. IBM developed it and maintains it. Experiment Impact Tracker draws emissions data from ElectricityMaps.org and in California it supports realtime carbon emission information by querying caiso.com.
Intel Performance Counter Monitor
Chip energy tracking tool
Intel Performance Counter Monitor is an API with tools to monitor performance and energy metrics of Intel Core, Xeon, Atom and Xeon Phi processors. It is an open-source tool developed and published by Intel.
PowerAPI
Software power tracking API
Step CI
Software missions tracking middleware
Step CI testing framework can add carbon emissions tracking to APIs that perform testing and quality assurance for software applications. It is an open-source API test automation framework that includes special co2 check. It comes from a French organization that specializes in sustainable IT and digital transformation services.
Grid Intensity Go
Job scheduling tool
Grid Intentisity Go is a tool to be written into software schedulers to enable grid carbon intensity to be factored into decisions about where and when to run jobs in the cloud. It is from the Green Web Foundation.
3. Website emissions calculators
This list is of online services that check the emissions impact of a website. It is useful for website owners to evaluate and improve their emissions. They can also be interesting to website visitors. Most suggest improvements to minimize emissions such as making pages smaller, removing unnecessary code, and using green web hosting.
Name
Type
Description
URLs
Are My Third Parties Green?
Are My Third Parties Green tests the emissions of third-party services used by a website. It is from The Green Web Foundation.
EcoGrader
EcoGrader measures carbon emissions from accessing a website and includes recommendations for reducing carbon emissions. From MightyBytes, a website design firm run by green software leader Tim Frick, Eco Grader uses information from The Green Web Foundation and Lighthouse. Note that the Green Web Foundation green hosting list includes companies that do not meet criteria of direct or self-generated low-carbon electricity” and does include companies that use offsets or unbundled RECs to claim carbon neutrality.
Green Web Foundation Calculator
The Green Web Foundation Calculator measures carbon emissions from accessing a website and includes recommendation to website owners for reducing carbon emissions. Note that the Green Web Foundation green hosting list includes companies that do not meet criteria of direct or self-generated low-carbon electricity” and does include companies that use offsets or unbundled RECs to claim carbon neutrality.
SitiGreen
SitiGreen (“Green Websites”) is for Italian websites, measures carbon emissions from accessing a website (based on CO2.js above).
Website Carbon Calculator
Website Carbon Calculator measures carbon emissions from accessing a website and raises awareness of the carbon footprint of the internet. From Wholegrain Digital.
EcoIndex.Fr
EcoIndex.fr is a volunteer-operated site that measures carbon emissions from accessing a website originially from GreenIT.fr. It runs on Neutral IT, a French web hosting company that is carbon negative with data centers under 1.06 PUE.
GreenIT-Analysis
Browser use emissions tracker
GreenIT-Analysis is a browser extension that tracks the environmental impacts of an entire user journey.
Referenced on this page
- Climatiq blog, “Measuring greenhouse gas emissions in data centres: the environmental impact of cloud computing”, May 24, 2023, https://www.climatiq.io/blog/measure-greenhouse-gas-emissions-carbon-data-centres-cloud-computing
- Green Software Foundation, “Why we joined the Real Time Cloud Carbon Footprint working group in the Green Software Foundation”, April 24, 2024, Ross Fairbanks, https://www.thegreenwebfoundation.org/news/why-we-joined-the-real-time-cloud-carbon-footprint-working-group-in-the-green-software-foundation
- Green Software Foundation, “Awesome Green Software” list https://github.com/Green-Software-Foundation/awesome-green-software
- The Green Web Foundation blog, “Why We Joined the Real Time Cloud Footprint Working Group in the Green Software Foundation”, Ross Fairbanks, April 23, 2024, https://www.thegreenwebfoundation.org/news/why-we-joined-the-real-time-cloud-carbon-footprint-working-group-in-the-green-software-foundation
- The Green Algorithms Project — Lannelongue, L., Grealey, J., Inouye, M., “Green Algorithms: Quantifying the Carbon Footprint of Computation”. Adv. Sci. 2021, 2100707. https://doi.org/10.1002/advs.202100707
- AWS Sustainability Blog, accessed July 2024, https://sustainability.aboutamazon.com/products-services/the-cloud
- Electricity Maps, https://www.electricitymaps.com
- Microsoft Azure Sustainability Blog, “Powering sustainable transformation”, accessed July 2024, https://datacenters.microsoft.com/globe/powering-sustainable-transformation
- Microsoft Azure Blog: “Choose the right Azure region for you” , accessed July 2024, https://datacenters.microsoft.com/globe/explore
- Google Cloud Blog, “Carbon free energy for Google Cloud regions”, accessed July 2024, https://cloud.google.com/sustainability/region-carbon
- Cloud Jewels blog on Etsy.com, “Cloud Jewels: Estimating kWh in the Cloud, April 23, 2020, Emily Sommer, Mike Adler, John Perkins, et.al., https://www.etsy.com/codeascraft/cloud-jewels-estimating-kwh-in-the-cloud/
- Green Software Foundation Projects, https://greensoftware.foundation/projects and https://greensoftwarefoundation.atlassian.net/wiki/spaces/~612dd45e45cd76006a84071a/pages