Laying the Foundation on OSS Environmental Sustainability

### Overview In this talk, Max Körbächer (co-chair of the CNCF Technical Advisory Group for Environmental Sustainability) outlines the role and early initiatives of the TAG, established under the Cloud Native Computing Foundation (CNCF). Highlighting the massive scale of open source software adoption and the substantial greenhouse gas emissions driven by information and communications technology (ICT)—comparable to aviation or construction—Max describes how the group works to bring sustainability and energy efficiency into cloud native projects and contributor mindsets.

### Key Topics - **Role and Boundaries of TAG Environmental Sustainability**: Founded to assist the CNCF Technical Oversight Committee, the group helps open source projects understand, measure, and optimize systems for environmental sustainability. Boundaries are clearly set: the TAG focuses strictly on software within cloud native technology rather than physical infrastructure (e.g., data center cooling, construction), external compliance, or formal audits. - **Awareness and Ecosystem Collaboration**: The group maintains an OSS landscape document, develops Kubernetes best practices, plans expansions to the Cloud Native Maturity Model, and coordinates global initiatives like the Cloud Native Sustainability Week alongside two working groups: Green Review and Comms. - **Software Carbon Intensity (SCI) Specification**: Developed by the Green Software Foundation (GSF), the SCI framework provides a standardized mathematical approach to calculate emissions, considering operational energy, embodied carbon (unit M), and functional scaling factors (unit R). - **Software vs. Project Footprints**: Measuring a single packaged software artifact across idle, loaded, and Pareto-optimal states is significantly more tractable than calculating the entire software supply chain (PRs, CI/CD pipelines, container registries, and developer hardware).

### Technical Details - **Measurement Architecture & Pipeline**: The TAG's Green Review working group is implementing a reproducible testing pipeline. Project repositories provide deployment and load testing manifests along with an SCI specification file. The TAG runs these tests in standardized bare-metal environments to ensure comparability. - **Tooling & Observability**: Power consumption at the container and service level is captured using tools like Kepler (which leverages machine learning models to estimate energy mix and power per namespace/application) or Scaphandre on virtual machines. Metrics are exported to CNCF’s DevStats server to track SCI scores across software releases. - **Impact Engine Framework (IEF)**: Utilizing the GSF Impact Engine Framework and `ief.yaml` manifests, complex distributed applications (e.g., multi-cloud setups spanning Azure, AWS, GCP) can define calculators, map components, and export clear pipeline execution metrics. - **Challenges**: Key obstacles include the paradox of test suites consuming extra power, paywalled carbon energy mix data (e.g., WattTime, Electricity Maps), legal constraints restricting contributor commits to external repos, and the mathematical complexity of apportioning embodied carbon ($M$) across server lifespans, geographic grid carbon intensities, and partial server utilization rates.

### Takeaways Cloud native sustainability requires incremental, transparent actions rather than complete rewrites. By adopting standard frameworks like SCI, leveraging measurement tools such as Kepler and Scaphandre, and tracking standardized release metrics on DevStats, open source projects can gain visibility into their energy demands and make informed architectural optimizations.