ContextQMD
Back to Gitlabhq

CI/CD sustainability

doc/ci/sustainability/_index.md

19.2.03.4 KB
Original Source

{{< details >}}

  • Tier: Free, Premium, Ultimate
  • Offering: GitLab.com, GitLab Self-Managed, GitLab Dedicated

{{< /details >}}

[!note] The sustainability tools described on this page are third-party integrations. GitLab does not maintain or provide support for these tools, and makes no representation that these tools satisfy any regulatory or compliance requirements.

You can integrate third-party tools to measure, analyze, and reduce the carbon emissions from your pipelines and infrastructure.

The right tool depends on what you want to measure:

  • To measure emissions from CI/CD pipeline jobs, use Eco CI.
  • To measure emissions from virtual machines, pods, or application workloads, use Carmen.

Eco CI

Eco CI measures energy consumption and carbon emissions of CI/CD pipelines. It runs as lightweight bash scripts in your pipeline jobs and does not require separate servers or databases.

Use Eco CI to:

  • Measure the energy consumption and carbon emissions of CI/CD pipeline jobs.
  • Identify resource-intensive jobs in your pipeline.
  • Track carbon emissions per pipeline run over time.
  • Display a carbon emissions badge in your project README.

Carmen

Carmen (Carbon Measurement Engine) is an open source tool that measures carbon emissions from virtual machines, pods, and application workloads. It is built on the Green Software Foundation Impact Framework.

Use Carmen to:

  • Identify which services, VMs, or pods in your stack emit the most CO2.
  • Compare components and prioritize reductions.
  • Feed per-component carbon scores into your own FinOps dashboards or internal tooling.

Do not use Carmen for:

  • Corporate or ESG reporting.
  • Press releases or marketing material.
  • Compliance or regulatory disclosures.

Best practices

The following strategies can help reduce the carbon emissions of your CI/CD workflows.

Optimize job execution

To optimize job execution:

  • Use caching to avoid redundant work.
  • Run resource-intensive builds once in an early job, then share the output as an artifact with later jobs that need it.
  • Set appropriate timeout values to prevent runaway jobs.
  • Use smaller Docker images to reduce download and startup time.

Choose efficient runners

To choose efficient runners:

  • Select runner instance types that match your workload requirements.
  • Avoid over-provisioning resources for small jobs.
  • Use spot instances for non-critical workloads.
  • Use autoscaling to match capacity with demand.

Schedule strategically

To schedule strategically:

  • Schedule resource-intensive pipelines to run when most renewable energy is available in your CI server's region. Check Electricity Maps to find the best times and regions. Midday is usually a good default choice.
  • Consider carbon-aware scheduling for non-urgent pipelines.
  • Batch similar jobs together to improve resource utilization.

Monitor and iterate

To monitor and iterate on your sustainability efforts:

  • Establish baseline metrics for your pipelines.
  • Set targets for emission reduction.
  • Review high-impact jobs regularly for optimization opportunities.
  • Share sustainability metrics with your team.