Accessing resources from Microsoft Azure

This document shows you how to use identity federation to access Google Cloud resources from Microsoft Azure.

Traditionally, applications running outside Google Cloud have used service account keys to access Google Cloud resources. Using identity federation, you can let a managed identity for an Azure resource impersonate a service account. This lets your workload access Google Cloud resources directly, using a short-lived access token, and eliminates the maintenance and security burden associated with service account keys.

Before you begin

  1. Enable the IAM, Resource Manager, Service Account Credentials, and Security Token Service (STS) APIs.

    Enable the APIs

  2. Ensure you have the Workload Identity Pool Admin role (roles/iam.workloadIdentityPoolAdmin).

    Alternatively, the IAM Owner (roles/owner) basic role also includes permissions to configure identity federation. You should not grant basic roles in a production environment, but you can grant them in a development or test environment.

  3. Update the organization policy for your organization to allow federation from Azure.

  4. Create a Google Cloud service account.

  5. Grant the service account access to call the Google Cloud APIs required by your workload.

Creating a workload identity pool

You can use a workload identity pool to organize and manage external identities. Workload identity pools are isolated from each other, but a single pool can impersonate any number of service accounts. In general, we recommend creating a new pool for each of your environments, such as development, staging, or production.

To create a new workload identity pool, you'll need to provide an ID. You can also provide an optional description and display name. The ID cannot begin with gcp-; this prefix is reserved for use by Google.

gcloud

Execute the gcloud iam workload-identity-pools create command to create a workload identity pool:

gcloud iam workload-identity-pools create pool-id \
    --location="global" \
    --description="description" \
    --display-name="display-name"

The response looks like:

Created workload identity pool [pool-id].

REST

The projects.locations.workloadIdentityPools.create method creates a workload identity pool.

HTTP method and URL:

POST https://iam.googleapis.com/v1/projects/project-id/locations/global/workloadIdentityPools?workloadIdentityPoolId=pool-id

Request JSON body:

{
  "description": "description",
  "display-name": "display-name"
}

To send your request, expand one of these options:

The method returns a long-running Operation similar to the following:

{
  "name": "projects/project-number/locations/global/workloadIdentityPools/pool-id/operations/operation-id"
}

Adding Azure as an identity provider

To configure Azure as an identity provider for your workload identity pool, supply at least the following:

  • An ID for the provider.

  • The workload identity pool ID from the previous section in this document.

  • Your Azure Tenant ID.

  • A list of attribute mappings that map the claims on a token for an Azure-managed identity to the attributes on a Google token. Use assertion to refer to the Azure token, google for Google attributes, and attribute for custom attributes.

    There are two Google attributes: google.subject and google.groups. You can reference these attributes in IAM role bindings. google.subject also appears in Cloud Logging log entries.

    You must provide a mapping for google.subject. In general, we recommend mapping it to assertion.sub, which contains the object ID of a managed identity that you'll create in the next section. This provides a stable identifier for use in IAM role bindings. The mapping looks like this:

    google.subject=assertion.sub
    

    For more complex assertions, you can use the Common Expression Language. For example, if your workload identity pool contains multiple identity providers, you can append a prefix to disambiguate between them:

    google.subject="azure::" + assertion.tid + "::" + assertion.sub
    

    The google.subject field cannot exceed 127 characters.

    You can also specify custom attributes. For example, the following maps assertion.tid to attribute.tid:

    attribute.tid=assertion.tid
    

    The following example assigns a display name based on the value of assertion.oid:

    attribute.managed_identity_name={
    "8bb39bdb-1cc5-4447-b7db-a19e920eb111":"workload1",
    "55d36609-9bcf-48e0-a366-a3cf19027d2a":"workload2"
    }[assertion.oid]
    

    To obtain a complete list of claims you can reference, get an access token for an Azure VM in your workload. In your request, replace the resource parameter with the full resource name of your workload identity pool.

    For example:

    curl

    curl -s \
      'http://169.254.169.254/metadata/identity/oauth2/token?api-version=2018-02-01&resource=https://iam.googleapis.com/projects/project-number/locations/global/workloadIdentityPools/pool-id/providers/provider-id&object_id=managed-identity-object-id' \
      -H Metadata:true -H "Cache-Control: no-cache"
    

    PowerShell

    Invoke-WebRequest \
        -Uri 'http://169.254.169.254/metadata/identity/oauth2/token?api-version=2018-02-01&resource=https://iam.googleapis.com/projects/project-number/locations/global/workloadIdentityPools/pool-id/providers/provider-id&object_id=managed-identity-object-id' \
        -Headers @{Metadata="true"}
    

    The response is a JSON object with an access_token field, which contains an access token for the Azure VM. To decode the access token and view the available claims, follow these steps:

    1. Copy the entire access token.
    2. In a web browser, go to https://jwt.ms/.
    3. Paste the access token into the text box.
    4. Click Claims.

    To refer to a specific part of a claim in an expression, use the CEL extract() function, which extracts a value from a claim based on a template you provide. To learn more about extract(), see Extracting values from attributes.

    To check if a credential contains an claim, use the has() function.

You can also provide several optional parameters:

  • A display name and description.

  • A condition specifying attributes that the principal must present. The condition can apply to claims on the Azure credential, or attributes on the Google credential. Any request that does not meet the condition is rejected.

    Conditions are formatted as a CEL expression that returns a boolean. For example, the following rejects requests from any identity that isn't a member of a specific group:

    "e968c2ef-047c-498d-8d79-16ca1b61e77e" in assertion.groups
    

    To learn more about common use cases for conditions, see the workload identity federation overview.

The following example demonstrates adding Azure as an identity provider:

gcloud

Execute the gcloud iam workload-identity-pools providers create-oidc command to add Azure as an identity provider:

gcloud iam workload-identity-pools providers create-oidc provider-id \
    --workload-identity-pool="pool-id" \
    --issuer-uri="https://sts.windows.net/azure-tenant-id" \
    --location="global" \
    --attribute-mapping="google.subject=assertion.sub"

The response looks like:

Created workload identity pool provider [provider-id].

REST

The projects.locations.workloadIdentityPools.providers.create method adds Azure as a provider.

HTTP method and URL:

POST https://iam.googleapis.com/v1/projects/project-id/locations/global/workloadIdentityPools/pool-id/providers?workloadIdentityPoolProviderId=provider-id

Request JSON body:

{
  "attributeMapping": {
    "google.subject": "assertion.sub"
  },
  "oidc": {
    "issuerUri": "https://sts.windows.net/azure-tenant-id"
  }
}

To send your request, expand one of these options:

The method returns a long-running Operation similar to the following:

{
  "name": "projects/project-number/locations/global/workloadIdentityPools/pool-id/providers/provider-id/operations/operation-id"
}

Configuring your Azure tenant for identity federation

To prepare your Azure tenant for identity federation:

  1. Create an Azure AD application and service principal, and set its Application ID URI to the full resource name of the provider you created in the previous section:

    https://iam.googleapis.com/projects/project-number/locations/global/workloadIdentityPools/pool-id/providers/provider-id
    
  2. Create a managed identity, and take note of its object ID.

  3. Assign the managed identity to a virtual machine that you want to grant access to Google Cloud resources.

Giving permission to impersonate a service account

External identities can't access most Google Cloud resources directly. Instead, you let the identities impersonate a service account by granting them the Workload Identity User role (roles/iam.workloadIdentityUser).

To add this role binding for a specific managed identity:

gcloud iam service-accounts add-iam-policy-binding service-account-email \
    --role roles/iam.workloadIdentityUser \
    --member "principal://iam.googleapis.com/projects/project-number/locations/global/workloadIdentityPools/pool-id/subject/managed-identity-object-id"

To add this binding for all identities in a pool:

gcloud iam service-accounts add-iam-policy-binding service-account-email \
    --role roles/iam.workloadIdentityUser \
    --member "principalSet://iam.googleapis.com/projects/project-number/locations/global/workloadIdentityPools/pool-id/group/azure-tenant-id"

You can also grant access based on custom attributes. For example:

gcloud iam service-accounts add-iam-policy-binding service-account-email \
    --role="roles/iam.workloadIdentityUser" \
    --member="principalSet://iam.googleapis.com/projects/project-number/locations/global/workloadIdentityPools/pool-id/attribute.custom-attribute-name/custom-attribute-value"

To revoke access, replace add-iam-policy-binding with remove-iam-policy-binding.

You can also add or revoke bindings using the REST API or client libraries. To learn more, see Granting, changing, and revoking access to resources.

Generating Google credentials

If you use a supported client library, you can configure the client library so that it generates Google credentials automatically. Alternatively, you can generate Azure credentials manually, then exchange them for Google credentials.

When possible, we recommend that you generate credentials automatically, so that you do not need to implement the token-exchange process yourself.

Automatically generating credentials

If you access Google Cloud with a client library for one of the following languages, you can configure the client library to automatically generate credentials by using identity federation:

C++

Most of the Google Cloud Client Libraries for C++ support identity federation by using a ChannelCredentials object, which is created by calling grpc::GoogleDefaultCredentials(). To initialize this credential, you must build the client libraries with version 1.36.0 or later of gRPC.

The Cloud Storage Client Library for C++ uses the REST API, not gRPC, so it does not support identity federation.

Go

Client libraries for Go support identity federation if they use version v0.0.0-20210218202405-ba52d332ba99 or later of the golang.org/x/oauth2 module.

To check which version of this module your client library uses, run the following commands:

cd $GOPATH/src/cloud.google.com/go
go list -m golang.org/x/oauth2

Java

Client libraries for Java support identity federation if they use version 0.24.0 or later of the com.google.auth:google-auth-library-oauth2-http artifact.

To check which version of this artifact your client library uses, run the following Maven command in your application directory:

mvn dependency:list -DincludeArtifactIds=google-auth-library-oauth2-http

Node.js

Client libraries for Node.js support identity federation if they use version 7.0.2 or later of the google-auth-library package.

To check which version of this package your client library uses, run the following command in your application directory:

npm list google-auth-library

When you create a GoogleAuth object, you can specify a project ID, or you can allow GoogleAuth to find the project ID automatically. To find the project ID automatically, the service account in the configuration file must have the Browser role (roles/browser), or a role with equivalent permissions, on your project. For details, see the README for the google-auth-library package.

Python

Client libraries for Python support identity federation if they use version 1.27.0 or later of the google-auth package.

To check which version of this package your client library uses, run the following command in the environment where the package is installed:

pip show google-auth

To specify a project ID for the authentication client, you can set the GOOGLE_CLOUD_PROJECT environment variable, or you can allow the client to find the project ID automatically. To find the project ID automatically, the service account in the configuration file must have the Browser role (roles/browser), or a role with equivalent permissions, on your project. For details, see the user guide for the google-auth package.

To configure the client library to generate credentials automatically, run the gcloud iam workload-identity-pools create-cred-config command to generate a JSON configuration file:

gcloud iam workload-identity-pools create-cred-config \
    projects/project-number/locations/global/workloadIdentityPools/pool-id/providers/provider-id \
    --service-account=service-account-email \
    --output-file=filepath \
    --azure

Replace the following values:

  • project-number: The numeric ID for the project.
  • pool-id: The ID for the workload identity pool.
  • provider-id: The ID for the workload identity pool provider.
  • service-account-email: The email address of the service account to impersonate.
  • filepath: The filepath for the configuration file.

After you generate the configuration file, set the environment variable GOOGLE_APPLICATION_CREDENTIALS to the filepath for the configuration file. This environment variable tells the client library to use Application Default Credentials to authenticate. For details, see Finding credentials automatically.

Manually exchanging credentials

Once your Azure-managed identity has the ability to impersonate a service account, you can manually exchange its credentials for Google credentials.

To exchange credentials:

  1. Use the Azure Instance Metadata Service (IMDS) to obtain an Azure access token.

    Set the resource query parameter to the following value:

    https://iam.googleapis.com/projects/project-number/locations/global/workloadIdentityPools/pool-id/providers/provider-id
    

    Set the object_id query parameter to the object ID for the managed identity that you created earlier.

  2. Pass the Azure access token to the Security Token Service token() method to get a federated access token:

    REST

    The token method exchanges a third-party token for a Google token.

    Before using any of the request data below, make the following replacements:

    • project-number: Your Google Cloud project number.
    • pool-id: The ID of the workload identity pool you created earlier in this tutorial.
    • provider-id: The ID of the identity provider you configured earlier in this tutorial.

    HTTP method and URL:

    POST https://sts.googleapis.com/v1/token

    Request JSON body:

    {
      "audience": "//iam.googleapis.com/projects/project-number/locations/global/workloadIdentityPools/pool-id/providers/provider-id",
      "grantType": "urn:ietf:params:oauth:grant-type:token-exchange",
      "requestedTokenType": "urn:ietf:params:oauth:token-type:access_token",
      "scope": "https://www.googleapis.com/auth/cloud-platform",
      "subjectTokenType": "urn:ietf:params:oauth:token-type:jwt",
      "subjectToken": "azure-id-token"
    }
    

    To send your request, expand one of these options:

     

    The method returns a federated token.

  3. Call generateAccessToken() to exchange the federated token for a service account access token. A limited number of Google Cloud APIs support federated tokens; all Google Cloud APIs support service account access tokens.

    REST

    The Service Account Credentials API's serviceAccounts.generateAccessToken method generates an OAuth 2.0 access token for a service account.

    Before using any of the request data below, make the following replacements:

    • PROJECT_ID: Your Google Cloud project ID. Project IDs are alphanumeric strings, like my-project.
    • SA_ID: The ID of your service account. This can either be the service account's email address in the form SA_NAME@PROJECT_ID.iam.gserviceaccount.com, or the service account's unique numeric ID.
    • token: The federated access token.

    HTTP method and URL:

    POST https://iamcredentials.googleapis.com/v1/projects/-/serviceAccounts/SA_NAME@PROJECT_ID.iam.gserviceaccount.com:generateAccessToken

    Request JSON body:

    {
      "scope": [
        "https://www.googleapis.com/auth/cloud-platform"
      ]
    }
    

    To send your request, expand one of these options:

    If the generateAccessToken request was successful, the response body contains an OAuth 2.0 access token and an expiration time. The accessToken can then be used to authenticate a request on behalf of the service account until the expireTime has been reached:

    {
      "accessToken": "eyJ0eXAi...NiJ9",
      "expireTime": "2020-04-07T15:01:23.045123456Z"
    }
    

Once you have an access token for a service account, you can use it to call Google Cloud APIs by including the token in the Authorization header of your requests:

Authorization: Bearer service-account-access-token

The request is authorized as the service account.

What's next