Control IAM access to resources based on data sensitivity

This page describes how to automatically grant or deny Identity and Access Management (IAM) access to resources based on the sensitivity of the data in those resources.

You can configure the Sensitive Data Protection discovery service to automatically attach tags to resources based on the calculated sensitivity levels of those resources. You can then use IAM conditions to grant or deny access to a resource, based on the presence or absence of a sensitivity level tag key or tag value.

Suppose that you want your organization's data team to be able to freely copy and share BigQuery data in their daily operations. However, you don't know if any of that data contains your customers' personally identifiable information (PII). You can run discovery to classify the sensitivity levels of your BigQuery data. Then, conditionally grant access to the data team, such that they are only able to access BigQuery tables that have only low-sensitivity data.

For more information about how Sensitive Data Protection calculates data sensitivity, see Data risk and sensitivity levels.

For more information about using tags to control access to resources, see Tags and conditional access in the IAM documentation. For general information about tags, see Tags overview in the Resource Manager documentation.

Sensitivity level tag

This document uses the term sensitivity level tag to refer to a tag that you automatically attach to a resource to indicate the calculated sensitivity level of data in that resource.

Benefits

This feature offers the following benefits:

  • Automates access control across various supported resources based on attributes and classifications of the data in those resources. Automation helps you keep up with the growth and changes in the data in your organization, folders, and projects.
  • Restricts access to the supported resources until those resources are profiled and classified by Sensitive Data Protection. This practice is in accordance with the secure by default principle.
  • You can configure discovery to update the sensitivity level tag value each time it profiles your data. Consequently, a principal's access to a resource changes automatically as the calculated data sensitivity level for that resource changes.
  • You can configure discovery to lower the calculated data risk level on a resource if discovery detects that a sensitivity level tag is present on that resource. This option helps you measure the improvement in your data security and privacy posture.

Supported resources

With this feature, Sensitive Data Protection automatically tags data at following levels:

  • BigQuery tables
  • Cloud SQL instances
  • Cloud Storage buckets

How it works

The following is a high-level workflow for controlling access to resources based on data sensitivity. These tasks don't have to be performed by the same person.

  1. Create a sensitivity level tag
  2. Grant conditional access to resources based on the value of the sensitivity level tag
  3. Enable the automatic tagging in the discovery configuration
  4. Grant the service agent permission to attach the sensitivity level tag to resources

Required permissions

The permissions that you need depend on the action that you need to perform.

To gain these permissions, ask your administrator to grant the suggested role at the appropriate level of the resource hierarchy.

Permissions for administering tags

See Administer tags in the Resource Manager documentation.

Permissions for granting conditional access to resources

See Required roles in the IAM documentation.

Permissions for configuring discovery

See Roles required to configure and view data profiles.

Create a sensitivity level tag

In this task, you create a tag key with tag values that map to the data sensitivity levels that Sensitive Data Protection uses to classify your data. For example, you can use the following tag key and tag values.

  1. Set the tag key to sensitivity-level.

  2. Set the following tag values:

    low
    Tag value to attach to low-sensitivity data
    moderate
    Tag value to attach to moderate-sensitivity data
    high
    Tag value to attach to high-sensitivity data

    You can also create a tag value for resources that have an unknown sensitivity level. Alternatively, you can reuse any of the low, medium, or high tag values for those resources.

For simplicity, this example sets a one-to-one mapping between a tag value and a sensitivity level. In practice, you can customize the tag values to suit your business needs. For example, you might use values like confidential, PII, or SPII (sensitive PII).

Sensitive data discovery can be configured at the organization, folder, and project levels. If you intend to use this sensitivity level tag for organization-level or folder-level discovery, we recommend that you create this tag at the organization level.

For information about how to create a tag, see Creating and managing tags in the Resource Manager documentation.

Grant conditional access to resources based on the value of the sensitivity level tag

In this task, you grant a role to a principal only if the sensitivity level tag attached to the resource has a specific value. For example, you can grant a principal access to only data with moderate and low tag values.

This section provides example conditions that are formatted for use with the condition editor. The condition editor provides a text-based interface to manually enter an expression using CEL syntax. For information about how to attach IAM conditions to role bindings, see Manage conditional role bindings in the IAM documentation.

These examples follow the tagging model defined in Create a sensitivity level tag on this page.

Grant principals access to only low-sensitivity data

In this example, you grant access to a resource if the resource contains only low-sensitivity data. You can also use this example to restrict all access to the resource until sensitive data discovery has run on that resource.

resource.matchTagId("tagKeys/SENSITIVITY_LEVEL_TAG_KEY",
"tagValues/TAG_VALUE_FOR_LOW_SENSITIVITY")

Replace the following:

  • SENSITIVITY_LEVEL_TAG_KEY: the numeric ID of the sensitivity level tag key that you created
  • TAG_VALUE_FOR_LOW_SENSITIVITY: the numeric ID of the tag value that you created for low-sensitivity data

Grant principals access to only moderate-sensitivity and low-sensitivity data

In this example, you grant access to a resource if the resource contains only moderate-sensitivity or low-sensitivity data. Note that there is an OR operator between the two conditions.

resource.matchTagId("tagKeys/SENSITIVITY_LEVEL_TAG_KEY", "tagValues/TAG_VALUE_FOR_LOW_SENSITIVITY") ||
resource.matchTagId("tagKeys/SENSITIVITY_LEVEL_TAG_KEY", "tagValues/TAG_VALUE_FOR_MODERATE_SENSITIVITY")

Replace the following:

  • SENSITIVITY_LEVEL_TAG_KEY: the numeric ID of the sensitivity level tag key that you created
  • TAG_VALUE_FOR_LOW_SENSITIVITY: the numeric ID of the tag value that you created for low-sensitivity data
  • TAG_VALUE_FOR_MODERATE_SENSITIVITY: the numeric ID of the tag value that you created for moderate-sensitivity data

Grant principals access only if the sensitivity level tag is present

This is useful, for example, if you want to define an organization policy that requires all IAM access to be conditional based on the presence of a sensitivity level tag. You can also use this example to restrict all access to the resource until sensitive data discovery has run on that resource.

resource.hasTagKeyId("tagKeys/SENSITIVITY_LEVEL_TAG_KEY")

Replace SENSITIVITY_LEVEL_TAG_KEY with the numeric ID of the sensitivity level tag key that you created.

Example deny policies

For information about how to form a deny policy for use with a resource tag, see Structure of a deny policy. For a list of supported permissions, see Permissions supported by deny policies.

Deny access if there is no sensitivity level tag

The following excerpt of a deny policy denies the bigquery.googleapis.com/tables.get permission if the resource doesn't have the sensitivity level tag.

  "rules": [
    {
      "denyRule": {
        "deniedPrincipals": [
          "principalSet://goog/group/data-team@example.com"
        ],
        "deniedPermissions": [
          "bigquery.googleapis.com/tables.get"
        ],
        "denialCondition": {
          "title": "Resource has no key",
          "expression": "!resource.hasTagKeyId("tagKeys/SENSITIVITY_LEVEL_TAG_KEY")"
        }
      }
    }
  ]

Replace SENSITIVITY_LEVEL_TAG_KEY with the numeric ID of the sensitivity level tag key that you created.

Deny access if moderate-sensitivity or high-sensitivity data is present

The following excerpt of a deny policy denies the bigquery.googleapis.com/tables.get permission if the resource has any moderate-sensitivity or high-sensitivity data.

  "rules": [
    {
      "denyRule": {
        "deniedPrincipals": [
          "principalSet://goog/group/data-team@example.com"
        ],
        "deniedPermissions": [
          "bigquery.googleapis.com/tables.get"
        ],
        "denialCondition": {
          "title": "Resource has moderate or high data sensitivity",
          "expression": "resource.matchTagId("tagKeys/SENSITIVITY_LEVEL_TAG_KEY", "tagValues/TAG_VALUE_FOR_MODERATE_SENSITIVITY") || resource.matchTagId("tagKeys/SENSITIVITY_LEVEL_TAG_KEY", "tagValues/TAG_VALUE_FOR_HIGH_SENSITIVITY")"
        }
      }
    }
  ]

Replace the following:

  • SENSITIVITY_LEVEL_TAG_KEY: the numeric ID of the sensitivity level tag key that you created
  • TAG_VALUE_FOR_MODERATE_SENSITIVITY: the numeric ID of the tag value that you created for moderate-sensitivity data
  • TAG_VALUE_FOR_HIGH_SENSITIVITY: the numeric ID of the tag value that you created for high-sensitivity data

Enable the automatic tagging in the discovery configuration

In this task, you enable the Tag resources action. This action instructs Sensitive Data Protection to automatically tag your data according to its calculated sensitivity level. You perform this task when you create or edit a discovery scan configuration.

To automatically tag a resource according to its calculated sensitivity level, follow these steps:

  1. Turn on the Tag resources option.

  2. For each sensitivity level (high, moderate, low, and unknown), do the following:

    1. In the Select scope list, select the resource where you created the sensitivity level tag key.
    2. For Key, select the sensitivity level tag key.
    3. For Value, select the tag value that you created for the given sensitivity level.

    If you skip a sensitivity level, no tag is attached for it.

  3. If you want Sensitive Data Protection to lower the data risk level of a resource when the sensitivity level tag is present, select When a tag is applied to a resource, lower the data risk of its profile to LOW. This option helps you measure the improvement in your data security and privacy posture.

  4. Select one or both of the following options:

    • Tag a resource when it is profiled for the first time.

    • Tag a resource when its profile is updated. Select this option if you want Sensitive Data Protection to overwrite the sensitivity level tag value on succeeding discovery runs. Consequently, a principal's access to a resource changes automatically as the calculated data sensitivity level for that resource increases or decreases.

      Don't select this option if you plan to manually update the sensitivity level tag values that the discovery service attached to your resources. If you select this option, Sensitive Data Protection can overwrite your manual updates.

Sensitive data discovery can be configured at the organization, folder, and project levels. If you intend to use this sensitivity level tag for organization-level discovery and if you don't want project-level discovery operations to overwrite the tag values that are set by the organization-level discovery, then make sure that only the service agent of the organization-level discovery configuration can attach this tag to resources.

Troubleshoot errors

This section describes errors that you might encounter when using this feature and how to resolve those errors.

Maximum number of tags exceeded

Each resource can have a maximum of 50 key-value pairs attached. Attempting to tag a resource that already has the maximum number of tags causes profile generation to fail. You get the following error:

The resource RESOURCE_NAME cannot be tagged because there are
too many existing tags bound to the resource. You can either disable automatic
tagging or delete at least one tag binding from the resource.

To resolve this issue, detach a tag from the resource. For more information, see the following:

Alternatively, disable the Tag resources action in your discovery scan configuration.

A tag value was deleted or renamed and discovery attempted to attach it to a resource

If a tag value for the sensitivity level tag is deleted or renamed and Sensitive Data Protection attempts to attach that tag value to a profiled resource, you get the following error:

Tag value TAG_VALUE not found, it has possibly been either deleted or renamed.

To resolve this issue, do one of the following:

  • If the tag was deleted, recreate the deleted tag value. Make sure that the recreated tag value matches the tag value that is referenced in your discovery scan configuration. For more information, see Create a sensitivity level tag on this page.
  • If the tag value was renamed, update the discovery scan configuration to use the new name of the tag value.

The service agent is missing permissions

If the service agent doesn't have the permissions that are required to attach the sensitivity level tag to the profiled resources, you get the following error:

The DLP service account SERVICE_AGENT_NAME is missing
permissions needed for attaching tags to resources. Check that the role
'resourcemanager.tagUser' is granted to the DLP service account.

To resolve this issue, follow these steps:

  1. Get the service agent ID that is associated with your discovery scan configuration:

    1. Go to the discovery scan configurations list.

      Go to discovery scan configurations

    2. Select your scan configuration.
    3. On the details page that opens, copy the service agent ID. This ID is in the format of an email address.

  2. Grant the service agent the Tag User role (roles/resourcemanager.tagUser) on the sensitivity level tag.

    Alternatively, you can grant a custom role on the sensitivity level tag. The custom role must have the resourcemanager.tagValues.get permission and resource-specific permissions for managing tag bindings. You need the resource-specific createTagBinding, deleteTagBinding, and listEffectiveTags permissions. For example, for BigQuery tables, you need the following:

    • resourcemanager.tagValues.get
    • bigquery.tables.createTagBinding
    • bigquery.tables.deleteTagBinding
    • bigquery.tables.listEffectiveTags

    For information about how to grant a role at the tag level, see Managing access to tags in the Resource Manager documentation.

What's next