2025.01.28

[seanmcilroy29]

2025.01.28 Agenda/Minutes

---

Time 0800 (PT) / 1600 (BST) - See the time in your timezone

• Chair – Adrian Cockcroft
• Chair - Pindy Bhullar (UBS)
• Convener – Sean Mcilroy (Linux Foundation)

---

Antitrust Policy

Joint Development Foundation meetings may involve participation by industry competitors, and the Joint Development Foundation intends to conduct all of its activities in accordance with applicable antitrust and competition laws. It is, therefore, extremely important that attendees adhere to meeting agendas and be aware of and not participate in any activities that are prohibited under applicable US state, federal or foreign antitrust and competition laws.

If you have questions about these matters, please contact your company counsel or counsel to the Joint Development Foundation, DLA Piper.

---

Recordings

WG agreed to record all Meetings. This meeting recording will be available until the next scheduled meeting.

---

Roll Call

Please add 'Attended' to this issue during the meeting to denote attendance.

Any untracked attendees will be added by the GSF team below:

• Full Name, Affiliation, (optional) GitHub username

---

Agenda

• Approve agenda
• Approve

---

PR Review

• Fix tiny typo in readme #74

---

Discussion

• Cloud PUE: Comparing AWS, Azure and GCP Global Regions - Feedback
• For future review - which datapoints from the CSRD in Europe would help enrich the dataset? #75
• Amazon Renewable Energy Projects Updates #78
• Azure PUE data alignment #27 (comment)
• Understanding the GSF RTC dataset? - See details below

Ref Material

• GSF Real-Time Cloud Renewable Energy Percentage spreadsheet
• RTC Cloud Region Impact Framework Documentation
• miro dataflow in detail

---

Next Steps

• Key areas for the project to focus on.

---

AOB

• Topics added during the meeting

---

Next Meeting

• 11 Feb

---

Action Items

• To be added

---

Understanding the GSF RTC dataset?

The “official" dataset

We have an ‘official’ dataset, which is listed below, on the main branch on GitHub here:

https://github.com/Green-Software-Foundation/real-time-cloud/blob/main/Cloud%20Region%20Metadata.csv

And for each column the guidance for what the column represents is as detailed in Cloud_Region_Metadata_specification.md:

https://github.com/Green-Software-Foundation/real-time-cloud/blob/main/Cloud_Region_Metadata_specification.md

The latest change to this main branch is this merged in PR here.

f59515e

We don’t have any formal versioning scheme, so while the guidance has a version number of 0.0.4, if I wanted to refer to refer to that dataset, I’d probably want to refer to commit f59515e, if I wanted to be explicit about the data I am using.

The “dev” dataset

We also have a “dev" dataset that is under active development, in the dev branch. In the working group we might have some rough consensus about it being broadly correct, but it doesn’t carry any weight as an official release.

So, we can’t say that there is any consensus (or at least lack of objections raised during the feedback period), the way I think we can with the “official” dataset, on there in branch, and specifically the state of the data as of commit f59515e.

If we refer to this “dev" dataset, when in doubt if we wanted to remove all ambiguity, about what we’re referring to we might refer to the specific commit - i.e. commit aa905b1, for the most recent signed off commit.

Licensing

The contents of this project are licensed under the MIT license. This license is permissive enough for someone to take this project and create a dataset that they publish under an open data license. For example, it would be valid to use the data in that CSV file, modify it, and publish it under this Open Data Commons Open Database License if need be—as long as the contents of that License.md file are included in any published works as well.

https://opendatacommons.org/licenses/odbl/summary/

[adrianco]

Attended

[PindyBhullar]

Attended

[seanmcilroy29]

Attended

Vincent Poncet - Google
Tim Smolcic - HSBC

[mrchrisadams]

Attended

[seanmcilroy29]

https://docs.google.com/spreadsheets/d/1BvI9G3HRR2FLmQnhEhCQQCpWntsvhD1o/edit?gid=429647841#gid=429647841
https://xbrl.efrag.org/e-esrs/esrs-set1-2023.html

[seanmcilroy29]

Adrian opens the meeting at

Summary Meeting Notes

The meeting focused on the progress and future steps for the GSF (Green Software Foundation) project. Key points included the need to finalise 2024 data, with Henry expected to provide updates soon. Chris discussed using GSF data for cross-cloud optimisation tools like Spare Cause, emphasising the importance of backward compatibility and maintaining column order. Adrian highlighted the challenges in comparing PUE (Power Usage Effectiveness) across cloud providers and the need for accurate data. The team also discussed leveraging electricity maps data and the Corporate Sustainability Reporting Directive (CSRD) for more detailed carbon intensity data. Future steps include refining predictive models and engaging with cloud providers for accurate data updates.

Minutes

Understanding GSF Data and Licensing
Chris explains the use of GSF data by an organization called Spare Cause for cross-cloud optimization and carbon intensity.
Chris discusses the need for a user guide and sanity-checking the understanding of publishing this information under an MIT license. Adrian emphasizes the importance of backward compatibility and maintaining compatibility in terms of columns and column order.

Use Cases and Data Utilization
Chris suggests using Spare Cause as a case study for the data to demonstrate its utility. Tim agrees on the necessity of example use cases to help individuals understand how to utilise the data. Adrian highlights the significance of documenting use cases and making them available on the site to promote their adoption. Chris discusses the potential for employing Spare Cause's tools to optimise carbon efficiency and pricing in cloud computing.

PUE Data and Cloud Provider Comparisons
Adrian discusses a story he published comparing PUE data from three cloud providers. Vincent mentions that public tables from vendors often lack all regions. Adrian explains the challenges of collecting data from all regions and the impact of new capacity on PUE. Adrian notes that Google has good data, AWS has published data for the last two years, and Azure has regressed in data quality.

AI PUE and Power Usage Efficiency
Tim queries the multiplier for AI PUE and the specifications of GPUs utilised by cloud providers. Adrian elucidates the power delivery system and the efficiency of modern data centre constructions. Vincent highlights the effects of location and climate on PUE as well as the trade-offs in water consumption. Adrian examines the latest technologies targeting a PUE of 1.08 and the significance of efficient power delivery systems.

AWS Power Purchase Agreements and Data Updates
Adrian updates the group on AWS's new power purchase agreements (PPAs) and mentions the lack of updated data. He emphasises the need to track and analyse AWS's data updates. Sean proposes a timeline for cleaning up the development branch and merging it into the main branch to ensure effective tracking and analysis of AWS's data updates. He suggests establishing a timeline for tidying up the dev branch and integrating it into the main branch.

Predicted Data and Future Estimates
Chris discusses the potential of using electricity maps data to provide annual grid intensity figures. Adrian confirms that a column for electricity maps data exists and needs to be populated. Chris mentions the Corporate Sustainability Reporting Directive (CSRD) and its potential influence on data disclosure. Vincent updates environmental reports and PPAs, incorporating innovative approaches such as nuclear and geothermal agreements.

GSG Protocol and Scope Two Reporting
Vincent mentions that the GSG protocol is under revision and that Google has two seats on the expert group. Adrian and Vincent discuss the potential for more granular and locally matched scope two reporting. Sean suggests documenting the process and contributing to the dev branch. Chris plans to share links to blog posts and update the group on using real-time carbon data.

Meeting Conclusion and Next Steps
Adrian intends to work on the development branch and tidy it up for the next meeting. Chris outlines his plan to initiate discussions with Spare Cause and furnish useful data for their tooling.

Action Items

• Refine the predictive model for 2024 and 2025 data.
• Ping Henry for the promised PUE data.
• Get the pull request for the dev branch updates cleaned up and merged.
• Explore using the Sparcost tool as a use case for the real-time cloud data set.
• Share links and information on the CSRD and its potential impact on data centre carbon intensity reporting.