Sustainability for Google Cloud

Overview

Google Cloud publishes your estimated carbon emissions through Carbon Footprint. Unlike AWS and Azure, there is no new permission to grant: the permission that unlocks carbon data, billing.accounts.getCarbonInformation, is already part of the Billing Account Viewer role assigned to OneLens when you connected Google Cloud.

What is needed instead is a one-time configuration step. Carbon Footprint has no query API, so the data has to be exported to BigQuery before anything can read it, in the same way the detailed usage cost export already works.

Two steps, not one

Creating the export only delivers data from that point forward. The transfer does not backfill by itself. To report on anything earlier than the month you set it up, schedule a backfill as described in step 2. Skipping it leaves you with no historical baseline.

Step 1: Create the export

You need Project Owner or Project Editor on the destination project, and Billing Account Viewer or Carbon Footprint Viewer on the billing account.

1

Open Carbon Footprint

In the Google Cloud console, go to Carbon Footprint, then select the Cloud Billing account you want to export from the Billing account menu.

2

Start the export

Click Data export to open the side panel, select the project the export should live in, then click Create export.

We recommend using the same project that already holds your billing_export dataset, so that carbon and cost data can be queried together.

3

Configure the transfer

On the BigQuery Data Transfer Service page, confirm Source is set to Google Cloud Carbon Footprint Exports, give the transfer a display name, and choose the destination dataset. An existing dataset works, or you can select Create new dataset.

Confirm the billing account ID shown under Data source details, then click Save and select Allow in the access request window.

The transfer creates a monthly partitioned table named carbon_footprint and refreshes it on the 15th of each month.

Step 2: Schedule the backfill

1

Open the transfer

Go to the details page of the transfer you just created.

2

Request the backfill

Click Schedule Backfill, then select Run for a date range.

3

Choose the full range

Set the start date to February 15, 2021, which is the earliest date Google accepts, and the end date to today. Click OK to request the backfill.

February 15, 2021 is the earliest selectable date because that run contains January 2021 data, which is the beginning of the available history.

Step 3: Send the export location to OneLens

OneLens cannot discover where you created the export, so this is the one thing we need back from you.

Send us two values:

• the project ID that holds the export

• the dataset ID you selected or created

The table name itself is always carbon_footprint, set by Google, so there is nothing to tell us about it.

These are the two values you chose in step 1. If you need to look them up again, open BigQuery in the console, go to Data transfers, and select the Carbon Footprint transfer; the destination dataset is shown on its details page. The same information is available from the CLI, where --transfer_location must match the region your dataset is in:

bq ls --transfer_config --transfer_location=us --project_id=<your-project-id>

Step 4: Grant access if the export is outside the billing project

If the export lives in the same project as your billing_export dataset, the BigQuery roles OneLens already holds cover it and there is nothing further to do.

If you placed it in a different project, grant the OneLens service account the BigQuery Data Viewer role on that dataset. It is the same service account address you used during your original Google Cloud onboarding.

You can confirm the carbon permission is in place at any time:

gcloud iam roles describe roles/billing.viewer \
  --format="value(includedPermissions)" | tr ';' '\n' | grep -i carbon

This returns billing.accounts.getCarbonInformation. Note the -i: the permission name is capitalised, so a case-sensitive search returns nothing and looks like a missing permission.

What you get in OneLens

	Detail
Granularity	Monthly
History	From January 2021, once backfilled
Scopes	Scope 1, scope 2 and scope 3, plus carbon offsets
Basis	Both location-based and market-based
Breakdown by	Project, service, region
Published	On the 15th of the following month

Good to know

• Carbon Footprint covers more than 80 Google Cloud services, but not all of them. Google names Apigee, AppSheet, Google Security Operations and Looker as not covered. If you use those, their emissions are absent from the total rather than shown as zero, and should be accounted for separately.

• Emissions are attributed to projects, not resources. OneLens allocates to cost centres using the project key, in the same way as cost.

• Offsets are reported separately. Google provides both the gross figure and a total after offsets, so you can disclose whichever your framework requires.

• The export needs no additional IAM to read new datasets created later in the same billing project, because the OneLens grants are made at project level.