This page provides an overview of Identity and Access Management (IAM) and its use with controlling access to buckets and objects in Cloud Storage. To learn about other ways of controlling access to buckets and objects, see Overview of Access Control.
This page focuses on aspects of IAM that are relevant specifically to Cloud Storage. For a detailed discussion of IAM and its features generally, see Identity and Access Management. In particular, see the Managing IAM Policies section.
IAM allows you to control who has access to the resources in your Google Cloud project. Resources include Cloud Storage buckets and objects stored within buckets, as well as other Google Cloud entities such as Compute Engine instances.
The set of access rules you apply to a resource is called an IAM policy. An IAM policy applied to your project defines the actions that users can take on all objects or buckets within your project. An IAM policy applied to a single bucket defines the actions that users can take on that specific bucket and objects within it.
For example, you can create an IAM policy for one of your buckets that gives one user administrative control of that bucket. Meanwhile, you can add another user to your project-wide IAM policy that gives that user the ability to view objects in any bucket of your project.
Members are the "who" of IAM. Members can be individual users, groups, domains, or even the public as a whole. Members are assigned roles, which grant members the ability to perform actions in Cloud Storage as well as Google Cloud more generally. Each role is a collection of one or more permissions. Permissions are the basic units of IAM: each permission allows you to perform a certain action.
For example, the
storage.objects.create permission allows you to create
objects. This permission is found in roles such as Storage Object Creator,
where it is the only permission, and Storage Object Admin, where many
permissions are bundled together. If you give the Storage Object Creator
role to a member for a specific bucket, they can only create objects in that
bucket. If you give another member the Storage Object Admin role for the
bucket, they can do additional tasks, such as deleting objects, but still only
within that bucket. If these two users are assigned these roles and no others,
they won't have knowledge of any other buckets in your project.
If you give a third member the Storage Object Admin role, but do
so for your project and not just a single bucket, they have access to any
object in any bucket of your project.
Using IAM with Cloud Storage makes it easy to limit a member's permissions without having to modify each bucket or object permission individually.
Permissions allow members to perform specific actions on buckets or objects
in Cloud Storage. For example, the
allows a member to list the buckets in your project. You don't directly give
members permissions; instead, you assign roles, which have one
or more permissions bundled within them.
For a reference list of the IAM permissions that apply to Cloud Storage, see IAM Permissions for Cloud Storage.
Roles are a bundle of one or more permissions. For example, the
Storage Object Viewer role contains the permissions
storage.objects.list. You assign roles to members, which allows
them to perform actions on the buckets and objects in your project.
For a reference list of the IAM roles that apply to Cloud Storage, see IAM Roles for Cloud Storage.
Project-level roles vs. bucket-level roles
Granting roles at the bucket level does not affect any existing roles that you granted at the project level, and vice versa. Thus, you can use these two levels of granularity to customize your permissions. For example, say you want to give a user permission to read objects in any bucket but create objects only in one specific bucket. To achieve this, give the user the Storage Object Viewer role at the project level, thus allowing the user to read any object stored in any bucket in your project, and give the user the Storage Object Creator role at the bucket level for a specific bucket, thus allowing the user to create objects only in that bucket.
Some roles can be used at both the project level and the bucket level. When used at the project level, the permissions they contain apply to all buckets and objects in the project. When used at the bucket level, the permissions only apply to a specific bucket and the objects within it. Examples of such roles are Storage Admin, Storage Object Viewer, and Storage Object Creator.
Some roles can only be applied at one level. For example, you can only apply the Viewer role at the project level, while you can only apply the Storage Legacy Object Owner role at the bucket level.
Relation to ACLs
Legacy Bucket IAM roles work in tandem with bucket ACLs: when you add or remove a Legacy Bucket role, the ACLs associated with the bucket reflect your changes. Similarly, changing a bucket-specific ACL updates the corresponding Legacy Bucket role for the bucket.
|Legacy Bucket role||Equivalent ACL|
|Storage Legacy Bucket Reader||Bucket Reader|
|Storage Legacy Bucket Writer||Bucket Writer|
|Storage Legacy Bucket Owner||Bucket Owner|
All other bucket-level IAM roles, and all project-level IAM roles, work independently from ACLs. For example, if you give a user the Storage Object Viewer role, the ACLs remain unchanged. This means you can use bucket-level IAM roles to grant broad access to all objects within a bucket and use the fine-grained object ACLs to customize access to specific objects within the bucket.
IAM permission for changing ACLs
You can use IAM to grant members the permission needed to change
ACLs on objects. The following
storage.buckets permissions together allow
users to work with bucket ACLs and default object ACLs:
Similarly, the following
storage.objects permissions together allow users to
work with object ACLs:
While IAM has many predefined roles that cover common use cases, you may wish to define your own roles which contain bundles of permissions that you specify. To support this, IAM offers custom roles.
There are a number of different types of members. For example, Google accounts
and Google groups represent two general types, while
allUsers are two specialized types. For a list of typical
member types in IAM, see Concepts related to identity. In
addition to the types listed there, IAM supports the following
member types, which can be applied specifically to your Cloud Storage
bucket IAM policies:
PROJECT_ID is the ID of a specific project.
When you grant one of the above member types a role, all members with the
specified permission for the specified project get the role you select. For
example, say you want to give all members who have the Viewer role for the
my-example-project the Storage Object Creator role for one
of your buckets. To do so, give the member
the Storage Object Creator role for that bucket.
IAM Conditions allows you to set conditions that control how permissions are granted to members. Cloud Storage supports the following types of condition attributes:
resource.name: Grant access to buckets and objects based on the bucket or object name. You can also use
resource.typeto grant access to either buckets or objects, but doing so is mostly redundant with using
resource.name. The following example condition applies an IAM setting to all objects with the same prefix:
- Date/time: Set an expiration date for the
request.time < timestamp('2019-01-01T00:00:00Z')
These conditional expressions are logic statements that use a subset of the Common Expression Language (CEL). You specify conditions in the role bindings of a bucket's IAM policy.
Keep these restrictions in mind:
- You must enable uniform bucket-level access on a bucket before adding conditions at the bucket level. Although conditions are allowed at the project level, you should migrate all buckets in the project to uniform bucket-level access to prevent Cloud Storage ACLs from overriding project-level IAM conditions. You can apply a uniform bucket-level access constraint to enable uniform bucket-level access for all new buckets in your project.
gsutil iam chcommand does not work with policies that contain conditions. To modify a policy that has conditions, use
gsutil iam getto retrieve the policy for the relevant bucket, edit it locally, and then use
gsutil iam setto re-apply it to the bucket.
- gsutil must be at version 4.38 or higher to use conditions.
- When using the JSON API to call
setIamPolicyon buckets with conditions, you must set the IAM policy version to 3.
- Since the
storage.objects.listpermission is granted at the bucket level, you cannot use the
resource.namecondition attribute to restrict object listing access to a subset of objects in the bucket. Users without
storage.objects.listpermission at the bucket level can experience degraded functionality for the Console and gsutil.
- Expired conditions remain in your IAM policy until you remove them.
Using with Cloud Storage tools
Although IAM permissions cannot be set through the XML API, users granted IAM permissions can still use the XML API, as well as any other tool for accessing Cloud Storage.
IAM, like any other administrative settings, requires active management to be effective. Before you make a bucket or object accessible to other users, be sure you know who you want to share the bucket or object with and what roles you want each of those people to play. Over time, changes in project management, usage patterns, and organizational ownership may require you to modify IAM settings on buckets and projects, especially if you manage Cloud Storage in a large organization or for a large group of users. As you evaluate and plan your access control settings, keep the following best practices in mind:
Use the principle of least privilege when granting access.
The principle of least privilege is a security guideline for granting access to your resources. When you grant access based on the principle of least privilege, you give a user only the access they need to accomplish their assigned task. For example, if you want to share files with someone, you should grant them the
storage.objectReaderrole and not the
Avoid granting roles with
setIamPolicypermission to people you do not know.
setIamPolicypermission allows a user to change permissions and take control of data. You should use roles with
setIamPolicypermission only when you want to delegate administrative control over objects and buckets.
Be careful how you grant permissions for anonymous users.
allAuthenticatedUsersmember types should only be used when it is acceptable for anyone on the Internet to read and analyze your data. While these scopes are useful for some applications and scenarios, it is usually not a good idea to grant all users certain permissions such as
Understand the Viewer/Editor/Owner primitive roles in Cloud Storage
The Viewer/Editor/Owner primitive roles effectively grant the access their names imply within Cloud Storage; however they do so indirectly through additional access that is provided at the bucket and object levels, using member types unique to Cloud Storage. While the access is added by default, you can revoke it.
For example, the Viewer role allows members to list buckets, but not to view buckets or objects. However:
Members with the Viewer role typically obtain the Storage Legacy Bucket Reader role and its permissions for each bucket in the project. This is because when you create a new bucket, the default behavior for Cloud Storage is to grant the Storage Legacy Bucket Reader role for the new bucket to the Viewer primitive role. Consequently, all members who have the Viewer primitive role for the project also have the Storage Legacy Bucket Reader or the new bucket. This behavior allows members with the Viewer role to view buckets.
Buckets have a default object ACL of
projectPrivate, which means objects added to the bucket gain, by default, the
projectPrivateACL. This ACL allows members with the Viewer role to view objects.
Similarly, the Editor and Owner primitive roles have limited Cloud Storage access by themselves, but members assigned these roles gain Storage Legacy Bucket Owner for new buckets, and ownership of objects through the
Note that because some bucket and object access is not inherent to the Viewer/Editor/Owner primitive roles, it is possible to revoke access that members might otherwise expect they have.
Avoid setting permissions that result in inaccessible buckets and objects.
A bucket or object becomes inaccessible when there is no one with permission to read it. This can happen for a bucket when all IAM permissions on the bucket get removed. This can happen for an object when the object's owner leaves a project and there are no bucket or project-level IAM policies that grant access to objects. In both cases, you can regain access by assigning an appropriate role, such as Storage Admin, to yourself or another member at the project level. Note that doing so gives access to all buckets and objects in the project, so you may want to revoke the role once you've reset access to the affected bucket or object.
Be sure you delegate administrative control of your buckets.
By default, members with the project-level Owner role are the only entities that have the Storage Legacy Bucket Owner role on a newly created bucket. You should have at least two members with the Owner role at any given time so that if a team member leaves the group, your buckets can still be managed by the other members.