Using identity-based policies (IAM policies) for Amazon Redshift

This topic provides examples of identity-based policies in which an account administrator can attach permissions policies to IAM identities (that is, users, groups, and roles).

The following shows an example of a permissions policy. The policy allows a user to create, delete, modify, and reboot all clusters, and then denies permission to delete or modify any clusters where the cluster identifier starts with production in AWS Region us-west-2 and AWS account 123456789012.

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Sid":"AllowClusterManagement",
      "Action": [
        "redshift:CreateCluster",
        "redshift:DeleteCluster",
        "redshift:ModifyCluster",
        "redshift:RebootCluster"
      ],
      "Resource": [
        "*"
      ],
      "Effect": "Allow"
    },
    {
      "Sid":"DenyDeleteModifyProtected",
      "Action": [
        "redshift:DeleteCluster",
        "redshift:ModifyCluster"
      ],
      "Resource": [
        "arn:aws:redshift:us-west-2:123456789012:cluster:production*"
      ],
      "Effect": "Deny"
    }
  ]
}

The policy has two statements:

AWS managed policies for Amazon Redshift

AWS addresses many common use cases by providing standalone IAM policies that are created and administered by AWS. Managed policies grant necessary permissions for common use cases so you can avoid having to investigate what permissions are needed. For more information, see AWS managed policies in the IAM User Guide.

You can also create your own custom IAM policies to allow permissions for Amazon Redshift API operations and resources. You can attach these custom policies to the IAM roles or groups that require those permissions.

The following sections describe AWS managed policies, which you can attach to users in your account, and are specific to Amazon Redshift.

AmazonRedshiftReadOnlyAccess

Grants read-only access to all Amazon Redshift resources for an AWS account.

You can find the AmazonRedshiftReadOnlyAccess policy on the IAM console and AmazonRedshiftReadOnlyAccess in the AWS Managed Policy Reference Guide.

AmazonRedshiftFullAccess

Grants full access to all Amazon Redshift resources for an AWS account. Additionally, this policy grants full access to all Amazon Redshift Serverless resources.

You can find the AmazonRedshiftFullAccess policy on the IAM console and AmazonRedshiftFullAccess in the AWS Managed Policy Reference Guide.

AmazonRedshiftQueryEditor

Grants full access to the query editor on the Amazon Redshift console.

You can find the AmazonRedshiftQueryEditor policy on the IAM console and AmazonRedshiftQueryEditor in the AWS Managed Policy Reference Guide.

AmazonRedshiftDataFullAccess

Grants full access to the Amazon Redshift Data API operations and resources for an AWS account.

You can find the AmazonRedshiftDataFullAccess policy on the IAM console and AmazonRedshiftDataFullAccess in the AWS Managed Policy Reference Guide.

AmazonRedshiftQueryEditorV2FullAccess

Grants full access to the Amazon Redshift query editor v2 operations and resources. This policy also grants access to other required services.

You can find the AmazonRedshiftQueryEditorV2FullAccess policy on the IAM console and AmazonRedshiftQueryEditorV2FullAccess in the AWS Managed Policy Reference Guide.

AmazonRedshiftQueryEditorV2NoSharing

Grants the ability to work with Amazon Redshift query editor v2 without sharing resources. This policy also grants access to other required services. The principal using this policy can't tag its resources (such as queries) to share them with other principals in the same AWS account.

You can find the AmazonRedshiftQueryEditorV2NoSharing policy on the IAM console and AmazonRedshiftQueryEditorV2NoSharing in the AWS Managed Policy Reference Guide.

AmazonRedshiftQueryEditorV2ReadSharing

Grants the ability to work with Amazon Redshift query editor v2 with limited sharing of resources. This policy also grants access to other required services. The principal using this policy can tag its resources (such as queries) to share them with other principals in the same AWS account. The granted principal can read the resources shared with its team but can't update them.

You can find the AmazonRedshiftQueryEditorV2ReadSharing policy on the IAM console and AmazonRedshiftQueryEditorV2ReadSharing in the AWS Managed Policy Reference Guide.

AmazonRedshiftQueryEditorV2ReadWriteSharing

Grants the ability to work with Amazon Redshift query editor v2 with sharing of resources. This policy also grants access to other required services. The principal using this policy can tag its resources (such as queries) to share them with other principals in the same AWS account. The granted principal can read and update the resources shared with its team.

You can find the AmazonRedshiftQueryEditorV2ReadWriteSharing policy on the IAM console and AmazonRedshiftQueryEditorV2ReadWriteSharing in the AWS Managed Policy Reference Guide.

AmazonRedshiftServiceLinkedRolePolicy

You can't attach AmazonRedshiftServiceLinkedRolePolicy to your IAM entities. This policy is attached to a service-linked role that allows Amazon Redshift to access account resources. For more information, see Using service-linked roles for Amazon Redshift.

You can find the AmazonRedshiftServiceLinkedRolePolicy policy on the IAM console and AmazonRedshiftServiceLinkedRolePolicy in the AWS Managed Policy Reference Guide.

AmazonRedshiftAllCommandsFullAccess

Grants the ability to use the IAM role created from the Amazon Redshift console and set it as default for the cluster to run the COPY from Amazon S3, UNLOAD, CREATE EXTERNAL SCHEMA, CREATE EXTERNAL FUNCTION, and CREATE MODEL commands. The policy also grants permissions to run SELECT statements for related services, such as Amazon S3, CloudWatch Logs, Amazon SageMaker AI, or AWS Glue.

You can find the AmazonRedshiftAllCommandsFullAccess policy on the IAM console and AmazonRedshiftAllCommandsFullAccess in the AWS Managed Policy Reference Guide.

You can also create your own custom IAM policies to allow permissions for Amazon Redshift API operations and resources. You can attach these custom policies to the IAM roles or groups that require those permissions.

Amazon Redshift updates to AWS managed policies

View details about updates to AWS managed policies for Amazon Redshift since this service began tracking these changes. For automatic alerts about changes to this page, subscribe to the RSS feed on the Amazon Redshift Document history page.

Permissions required to use Redshift Spectrum

Amazon Redshift Spectrum requires permissions to other AWS services to access resources. For details about permissions in IAM policies for Redshift Spectrum, see IAM policies for Amazon Redshift Spectrum in the Amazon Redshift Database Developer Guide.

Permissions required to use the Amazon Redshift console

For a user to work with the Amazon Redshift console, that user must have a minimum set of permissions that allows the user to describe the Amazon Redshift resources for their AWS account. These permissions must also allow the user to describe other related information, including Amazon EC2 security, Amazon CloudWatch, Amazon SNS, and network information.

If you create an IAM policy that is more restrictive than the minimum required permissions, the console doesn't function as intended for users with that IAM policy. To ensure that those users can still use the Amazon Redshift console, also attach the AmazonRedshiftReadOnlyAccess managed policy to the user. How to do this is described in AWS managed policies for Amazon Redshift.

For information to give a user access to the query editor on the Amazon Redshift console, see Permissions required to use the Amazon Redshift console query editor.

You don't need to allow minimum console permissions for users that are making calls only to the AWS CLI or the Amazon Redshift API.

Permissions required to use the Amazon Redshift console query editor

For a user to work with the Amazon Redshift query editor, that user must have a minimum set of permissions to Amazon Redshift and Amazon Redshift Data API operations. To connect to a database using a secret, you must also have Secrets Manager permissions.

To give a user access to the query editor on the Amazon Redshift console, attach the AmazonRedshiftQueryEditor and AmazonRedshiftReadOnlyAccess AWS managed policies. The AmazonRedshiftQueryEditor policy allows the user permission to retrieve the results of only their own SQL statements. That is, statements submitted by the same aws:userid as shown in this section of the AmazonRedshiftQueryEditor AWS managed policy.

{
    "Sid":"DataAPIIAMStatementPermissionsRestriction",
    "Action": [
        "redshift-data:GetStatementResult",
        "redshift-data:CancelStatement",
        "redshift-data:DescribeStatement",
        "redshift-data:ListStatements"
    ],
    "Effect": "Allow",
    "Resource": "*",
    "Condition": {
        "StringEquals": {
            "redshift-data:statement-owner-iam-userid": "${aws:userid}"
        }
    }
}

To allow a user to retrieve the results of SQL statements of others in the same IAM role, create your own policy without the condition to limit access to the current user. Also limit access to change a policy to an administrator.

Permissions required to use the query editor v2

For a user to work with the Amazon Redshift query editor v2, that user must have a minimum set of permissions to Amazon Redshift, the query editor v2 operations, and other AWS services such as AWS Key Management Service, AWS Secrets Manager, and tagging service.

To give a user full access to the query editor v2, attach the AmazonRedshiftQueryEditorV2FullAccess AWS managed policy. The AmazonRedshiftQueryEditorV2FullAccess policy allows the user permission to share query editor v2 resources, such as queries, with others in the same team. For details about how access to query editor v2 resources are controlled, see the definition of the specific managed policy for query editor v2 in the IAM console.

Some Amazon Redshift query editor v2 AWS managed policies use AWS tags within conditions to scope access to resources. Within query editor v2, sharing queries is based on the tag key and value "aws:ResourceTag/sqlworkbench-team": "${aws:PrincipalTag/sqlworkbench-team}" in the IAM policy attached to principal (the IAM role). Principals in the same AWS account with the same tag value (for example, accounting-team), are on the same team in query editor v2. You can only be associated with one team at a time. A user with administrative permissions can set up teams in the IAM console by giving all team members the same value for the sqlworkbench-team tag. If the tag value of the sqlworkbench-team is changed for an IAM user or an IAM role, there might be a delay until the change is reflected in shared resources. If the tag value of a resource (such as a query) is changed, again there might be a delay until the change is reflected. Team members must also have the tag:GetResources permission to share.

For more information on how to attach a tag to a principal, including IAM roles and IAM users, see Tagging IAM resources in the IAM User Guide.

You can also set up teams at the session level using an Identity Provider (IdP). This allows multiple users using the same IAM role to have different team. The IAM role trust policy must allow the sts:TagSession operation. For more information, see Permissions required to add session tags in the IAM User Guide. Add the principal tag attribute to the SAML assertion provided by your IdP.

<Attribute Name="https://aws.amazon.com/SAML/Attributes/PrincipalTag:sqlworkbench-team">
    <AttributeValue>accounting-team</AttributeValue>
</Attribute>

Follow the instructions for your Identity provider (IdP) to populate the SAML attribute with the content coming from your directory. For more information about Identity providers (IdPs) and Amazon Redshift, see Using IAM authentication to generate database user credentials and Identity providers and federation in the IAM User Guide.

The sqlworkbench:CreateNotebookVersion grants permission to get the current content of notebook cells and create a notebook version on your account. Meaning, at the time of version creation, the current content of the notebook is the same as the version’s content. Later on, the content of the cells in the version stay the same as the current notebook is updated. The sqlworkbench:GetNotebookVersion grants permission to get a version of the notebook. A user who doesn’t have sqlworkbench:BatchGetNotebookCell permission but has sqlworkbench:CreateNotebookVersion and sqlworkbench:GetNotebookVersion permissions on a notebook has access to notebook cells in the version. This user without the sqlworkbench:BatchGetNotebookCell permission is still able to retrieve the content of a notebook’s cells by first creating a version and then getting this created version.

Permissions required to use the Amazon Redshift scheduler

When you use the Amazon Redshift scheduler, you set up an IAM role with a trust relationship to the Amazon Redshift scheduler (scheduler.redshift.amazonaws.com) to allow the scheduler to assume permissions on your behalf. You also attach a policy (permissions) to the role for the Amazon Redshift API operations that you want to schedule.

The following example shows the policy document in JSON format to set up a trust relationship with the Amazon Redshift scheduler and Amazon Redshift.

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Principal": {
                "Service": [
                    "scheduler.redshift.amazonaws.com",
                    "redshift.amazonaws.com"
                ]
            },
            "Action": "sts:AssumeRole"
        }
    ]
}

For more information about trust entities, see Creating a role to delegate permissions to an AWS service in the IAM User Guide.

You also must add permission for the Amazon Redshift operations you want to schedule.

For the scheduler to use the ResizeCluster operation, add a permission that is similar to the following to your IAM policy. Depending on your environment, you might want to make the policy more restrictive.

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": "redshift:ResizeCluster",
            "Resource": "*"
        }
    ]
}

For the steps to create a role for the Amazon Redshift scheduler, see Creating a role for an AWS service (console) in the IAM User Guide. Make these choices when you create a role in the IAM console:

The IAM role you create has trusted entities of scheduler.redshift.amazonaws.com and redshift.amazonaws.com. It also has an attached policy that allows a supported Amazon Redshift API action, such as, "redshift:ResizeCluster".

Permissions required to use the Amazon EventBridge scheduler

When you use the Amazon EventBridge scheduler, you set up an IAM role with a trust relationship to the EventBridge scheduler (events.amazonaws.com) to allow the scheduler to assume permissions on your behalf. You also attach a policy (permissions) to the role for the Amazon Redshift Data API operations that you want to schedule and a policy for Amazon EventBridge operations.

You use the EventBridge scheduler when you create scheduled queries with the Amazon Redshift query editor on the console.

You can create an IAM role to run scheduled queries on the IAM console. In this IAM role, attach AmazonEventBridgeFullAccess and AmazonRedshiftDataFullAccess.

The following example shows the policy document in JSON format to set up a trust relationship with the EventBridge scheduler.

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Principal": {
                "Service": [
                    "events.amazonaws.com",
                ]
            },
            "Action": "sts:AssumeRole"
        }
    ]
}

For more information about trust entities, see Creating a role to delegate permissions to an AWS service in the IAM User Guide.

For the steps to create a role for the EventBridge scheduler, see Creating a role for an AWS service (console) in the IAM User Guide. Make these choices when you create a role in the IAM console:

The IAM role that you create has a trusted entity of events.amazonaws.com. It also has an attached policy that allows supported Amazon Redshift Data API actions, such as, "redshift-data:*".

Permissions required to use Amazon Redshift machine learning (ML)

Following, you can find a description of the permissions required to use Amazon Redshift machine learning (ML) for different use cases.

For your users to use Amazon Redshift ML with Amazon SageMaker AI, create an IAM role with a more restrictive policy than the default. You can use the policy following. You can also modify this policy to meet your needs.

The following policy shows the permissions required to run SageMaker AI Autopilot with model explainability from Amazon Redshift.

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": [
                "sagemaker:CreateTrainingJob",
                "sagemaker:CreateAutoMLJob",
                "sagemaker:CreateCompilationJob",
                "sagemaker:CreateEndpoint",
                "sagemaker:DescribeAutoMLJob",
                "sagemaker:DescribeTrainingJob",
                "sagemaker:DescribeCompilationJob",
                "sagemaker:DescribeProcessingJob",
                "sagemaker:DescribeTransformJob",
                "sagemaker:ListCandidatesForAutoMLJob",
                "sagemaker:StopAutoMLJob",
                "sagemaker:StopCompilationJob",
                "sagemaker:StopTrainingJob",
                "sagemaker:DescribeEndpoint",
                "sagemaker:InvokeEndpoint",
                "sagemaker:StopProcessingJob",
                "sagemaker:CreateModel",
                "sagemaker:CreateProcessingJob"
            ],
            "Resource": [
                "arn:aws:sagemaker:*:*:model/*redshift*",
                "arn:aws:sagemaker:*:*:training-job/*redshift*",
                "arn:aws:sagemaker:*:*:automl-job/*redshift*",
                "arn:aws:sagemaker:*:*:compilation-job/*redshift*",
                "arn:aws:sagemaker:*:*:processing-job/*redshift*",
                "arn:aws:sagemaker:*:*:transform-job/*redshift*",
                "arn:aws:sagemaker:*:*:endpoint/*redshift*"
            ]
        },
        {
            "Effect": "Allow",
            "Action": [
                "logs:CreateLogGroup",
                "logs:CreateLogStream",
                "logs:DescribeLogStreams",
                "logs:PutLogEvents"
            ],
            "Resource": [
                "arn:aws:logs:*:*:log-group:/aws/sagemaker/Endpoints/*redshift*",
                "arn:aws:logs:*:*:log-group:/aws/sagemaker/ProcessingJobs/*redshift*",
                "arn:aws:logs:*:*:log-group:/aws/sagemaker/TrainingJobs/*redshift*",
                "arn:aws:logs:*:*:log-group:/aws/sagemaker/TransformJobs/*redshift*"
            ]
        },
        {
            "Effect": "Allow",
            "Action": [
                "cloudwatch:PutMetricData"
            ],
            "Resource": "*",
            "Condition": {
                "StringEquals": {
                    "cloudwatch:namespace": [
                        "SageMaker",
                        "/aws/sagemaker/Endpoints",
                        "/aws/sagemaker/ProcessingJobs",
                        "/aws/sagemaker/TrainingJobs",
                        "/aws/sagemaker/TransformJobs"
                    ]
                }
            }
        },
        {
            "Effect": "Allow",
            "Action": [
                "ecr:BatchCheckLayerAvailability",
                "ecr:BatchGetImage",
                "ecr:GetAuthorizationToken",
                "ecr:GetDownloadUrlForLayer"
            ],
            "Resource": "*"
        },
        {
            "Effect": "Allow",
            "Action": [
                "s3:GetObject",
                "s3:GetBucketAcl",
                "s3:GetBucketCors",
                "s3:GetEncryptionConfiguration",
                "s3:GetBucketLocation",
                "s3:ListBucket",
                "s3:ListAllMyBuckets",
                "s3:ListMultipartUploadParts",
                "s3:ListBucketMultipartUploads",
                "s3:PutObject",
                "s3:PutBucketAcl",
                "s3:PutBucketCors",
                "s3:DeleteObject",
                "s3:AbortMultipartUpload",
                "s3:CreateBucket"
            ],
            "Resource": [
                "arn:aws:s3:::redshift-downloads",
                "arn:aws:s3:::redshift-downloads/*",
                "arn:aws:s3:::*redshift*",
                "arn:aws:s3:::*redshift*/*"
            ]
        },
        {
            "Effect": "Allow",
            "Action": [
                "s3:GetObject",
                "s3:GetBucketAcl",
                "s3:GetBucketCors",
                "s3:GetEncryptionConfiguration",
                "s3:GetBucketLocation",
                "s3:ListBucket",
                "s3:ListAllMyBuckets",
                "s3:ListMultipartUploadParts",
                "s3:ListBucketMultipartUploads",
                "s3:PutObject",
                "s3:PutBucketAcl",
                "s3:PutBucketCors",
                "s3:DeleteObject",
                "s3:AbortMultipartUpload",
                "s3:CreateBucket"
            ],
            "Resource": "*",
            "Condition": {
                "StringEqualsIgnoreCase": {
                    "s3:ExistingObjectTag/Redshift": "true"
                }
            }
        },
        {
            "Effect": "Allow",
            "Action": [
                "iam:PassRole"
            ],
            "Resource": "arn:aws:iam::*:role/*",
            "Condition": {
                "StringEquals": {
                    "iam:PassedToService": [
                        "redshift.amazonaws.com",
                        "sagemaker.amazonaws.com"
                    ]
                }
            }
        }
    ]
}

The following policy shows the full minimal permissions to allow access to Amazon DynamoDB, Redshift Spectrum and Amazon RDS federation.

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": [
                "sagemaker:CreateTrainingJob",
                "sagemaker:CreateAutoMLJob",
                "sagemaker:CreateCompilationJob",
                "sagemaker:CreateEndpoint",
                "sagemaker:DescribeAutoMLJob",
                "sagemaker:DescribeTrainingJob",
                "sagemaker:DescribeCompilationJob",
                "sagemaker:DescribeProcessingJob",
                "sagemaker:DescribeTransformJob",
                "sagemaker:ListCandidatesForAutoMLJob",
                "sagemaker:StopAutoMLJob",
                "sagemaker:StopCompilationJob",
                "sagemaker:StopTrainingJob",
                "sagemaker:DescribeEndpoint",
                "sagemaker:InvokeEndpoint",
                "sagemaker:StopProcessingJob",
                "sagemaker:CreateModel",
                "sagemaker:CreateProcessingJob"
            ],
            "Resource": [
                "arn:aws:sagemaker:*:*:model/*redshift*",
                "arn:aws:sagemaker:*:*:training-job/*redshift*",
                "arn:aws:sagemaker:*:*:automl-job/*redshift*",
                "arn:aws:sagemaker:*:*:compilation-job/*redshift*",
                "arn:aws:sagemaker:*:*:processing-job/*redshift*",
                "arn:aws:sagemaker:*:*:transform-job/*redshift*",
                "arn:aws:sagemaker:*:*:endpoint/*redshift*"
            ]
        },
        {
            "Effect": "Allow",
            "Action": [
                "logs:CreateLogGroup",
                "logs:CreateLogStream",
                "logs:DescribeLogStreams",
                "logs:PutLogEvents"
            ],
            "Resource": [
                "arn:aws:logs:*:*:log-group:/aws/sagemaker/Endpoints/*redshift*",
                "arn:aws:logs:*:*:log-group:/aws/sagemaker/ProcessingJobs/*redshift*",
                "arn:aws:logs:*:*:log-group:/aws/sagemaker/TrainingJobs/*redshift*",
                "arn:aws:logs:*:*:log-group:/aws/sagemaker/TransformJobs/*redshift*"
            ]
        },
        {
            "Effect": "Allow",
            "Action": [
                "cloudwatch:PutMetricData"
            ],
            "Resource": "*",
            "Condition": {
                "StringEquals": {
                    "cloudwatch:namespace": [
                        "SageMaker",
                        "/aws/sagemaker/Endpoints",
                        "/aws/sagemaker/ProcessingJobs",
                        "/aws/sagemaker/TrainingJobs",
                        "/aws/sagemaker/TransformJobs"
                    ]
                }
            }
        },
        {
            "Effect": "Allow",
            "Action": [
                "ecr:BatchCheckLayerAvailability",
                "ecr:BatchGetImage",
                "ecr:GetAuthorizationToken",
                "ecr:GetDownloadUrlForLayer"
            ],
            "Resource": "*"
        },
        {
            "Effect": "Allow",
            "Action": [
                "s3:GetObject",
                "s3:GetBucketAcl",
                "s3:GetBucketCors",
                "s3:GetEncryptionConfiguration",
                "s3:GetBucketLocation",
                "s3:ListBucket",
                "s3:ListAllMyBuckets",
                "s3:ListMultipartUploadParts",
                "s3:ListBucketMultipartUploads",
                "s3:PutObject",
                "s3:PutBucketAcl",
                "s3:PutBucketCors",
                "s3:DeleteObject",
                "s3:AbortMultipartUpload",
                "s3:CreateBucket"
            ],
            "Resource": [
                "arn:aws:s3:::redshift-downloads",
                "arn:aws:s3:::redshift-downloads/*",
                "arn:aws:s3:::*redshift*",
                "arn:aws:s3:::*redshift*/*"
            ]
        },
        {
            "Effect": "Allow",
            "Action": [
                "s3:GetObject",
                "s3:GetBucketAcl",
                "s3:GetBucketCors",
                "s3:GetEncryptionConfiguration",
                "s3:GetBucketLocation",
                "s3:ListBucket",
                "s3:ListAllMyBuckets",
                "s3:ListMultipartUploadParts",
                "s3:ListBucketMultipartUploads",
                "s3:PutObject",
                "s3:PutBucketAcl",
                "s3:PutBucketCors",
                "s3:DeleteObject",
                "s3:AbortMultipartUpload",
                "s3:CreateBucket"
            ],
            "Resource": "*",
            "Condition": {
                "StringEqualsIgnoreCase": {
                    "s3:ExistingObjectTag/Redshift": "true"
                }
            }
        },
        {
            "Effect": "Allow",
            "Action": [
                "dynamodb:Scan",
                "dynamodb:DescribeTable",
                "dynamodb:Getitem"
            ],
            "Resource": [
                "arn:aws:dynamodb:*:*:table/*redshift*",
                "arn:aws:dynamodb:*:*:table/*redshift*/index/*"
            ]
        },
        {
            "Effect": "Allow",
            "Action": [
                "elasticmapreduce:ListInstances"
            ],
            "Resource": [
                "arn:aws:elasticmapreduce:*:*:cluster/*redshift*"
            ]
        },
        {
            "Effect": "Allow",
            "Action": [
                "elasticmapreduce:ListInstances"
            ],
            "Resource": "*",
            "Condition": {
                "StringEqualsIgnoreCase": {
                    "elasticmapreduce:ResourceTag/Redshift": "true"
                }
            }
        },
        {
            "Effect": "Allow",
            "Action": [
                "lambda:InvokeFunction"
            ],
            "Resource": "arn:aws:lambda:*:*:function:*redshift*"
        },
        {
            "Effect": "Allow",
            "Action": [
                "glue:CreateDatabase",
                "glue:DeleteDatabase",
                "glue:GetDatabase",
                "glue:GetDatabases",
                "glue:UpdateDatabase",
                "glue:CreateTable",
                "glue:DeleteTable",
                "glue:BatchDeleteTable",
                "glue:UpdateTable",
                "glue:GetTable",
                "glue:GetTables",
                "glue:BatchCreatePartition",
                "glue:CreatePartition",
                "glue:DeletePartition",
                "glue:BatchDeletePartition",
                "glue:UpdatePartition",
                "glue:GetPartition",
                "glue:GetPartitions",
                "glue:BatchGetPartition"
            ],
            "Resource": [
                "arn:aws:glue:*:*:table/*redshift*/*",
                "arn:aws:glue:*:*:catalog",
                "arn:aws:glue:*:*:database/*redshift*"
            ]
        },
        {
            "Effect": "Allow",
            "Action": [
                "secretsmanager:GetResourcePolicy",
                "secretsmanager:GetSecretValue",
                "secretsmanager:DescribeSecret",
                "secretsmanager:ListSecretVersionIds"
            ],
            "Resource": [
                "arn:aws:secretsmanager:*:*:secret:*redshift*"
            ]
        },
        {
            "Effect": "Allow",
            "Action": [
                "secretsmanager:GetRandomPassword",
                "secretsmanager:ListSecrets"
            ],
            "Resource": "*",
            "Condition": {
                "StringEquals": {
                    "secretsmanager:ResourceTag/Redshift": "true"
                }
            }
        },
        {
            "Effect": "Allow",
            "Action": [
                "iam:PassRole"
            ],
            "Resource": "arn:aws:iam::*:role/*",
            "Condition": {
                "StringEquals": {
                    "iam:PassedToService": [
                        "redshift.amazonaws.com",
                        "glue.amazonaws.com",
                        "sagemaker.amazonaws.com",
                        "athena.amazonaws.com"
                    ]
                }
            }
        }
    ]
}

Optionally, to use a AWS KMS key for encryption, add the following permissions to the policy.

{
    "Effect": "Allow",
    "Action": [
        "kms:CreateGrant",
        "kms:Decrypt",
        "kms:DescribeKey",
        "kms:Encrypt",
        "kms:GenerateDataKey*"
    ],
    "Resource": [
        "arn:aws:kms:<your-region>:<your-account-id>:key/<your-kms-key>"
    ]
}

To allow Amazon Redshift and SageMaker AI to assume the preceding IAM role to interact with other services, add the following trust policy to the role.

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Principal": {
        "Service": [
          "redshift.amazonaws.com",
          "sagemaker.amazonaws.com",
          "forecast.amazonaws.com"
        ]
      },
      "Action": "sts:AssumeRole"
    }
  ]
}

In the preceding, the Amazon S3 bucket redshift-downloads/redshift-ml/ is the location where the sample data used for other steps and examples is stored. You can remove this bucket if you don't need to load data from Amazon S3. Or replace it with other Amazon S3 buckets that you use to load data into Amazon Redshift.

The your-account-id, your-role, and your-s3-bucket values are the account ID, role, and bucket that you specify in your CREATE MODEL command.

Optionally, you can use the AWS KMS keys section of the sample policy if you specify an AWS KMS key for use with Amazon Redshift ML. The your-kms-key value is the key that you use as part of your CREATE MODEL command.

When you specify a private virtual private cloud (VPC) for a hyperparameter tuning job, add the following permissions.

{
            "Effect": "Allow",
            "Action": [
            "ec2:CreateNetworkInterface",
            "ec2:CreateNetworkInterfacePermission",
            "ec2:DeleteNetworkInterface",
            "ec2:DeleteNetworkInterfacePermission",
            "ec2:DescribeNetworkInterfaces",
            "ec2:DescribeVpcs",
            "ec2:DescribeDhcpOptions",
            "ec2:DescribeSubnets",
            "ec2:DescribeSecurityGroups"
            ]
}

To work with model explanation, make sure that you have the permissions to call SageMaker AI API operations. We recommend that you use the AmazonSageMakerFullAccess managed policy. If you want to create an IAM role with a more restrictive policy, use the policy following.

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": [
        "sagemaker::CreateEndpoint",
        "sagemaker::CreateEndpointConfig",
        "sagemaker::DeleteEndpoint",
        "sagemaker::DeleteEndpointConfig",
        "sagemaker::DescribeEndpoint",
        "sagemaker::DescribeEndpointConfig",
        "sagemaker::DescribeModel",
        "sagemaker::InvokeEndpoint",
        "sagemaker::ListTags"
      ],
      "Resource": "*"
    }
  ]
}

For more information about the AmazonSageMakerFullAccess managed policy, see AmazonSageMakerFullAccess in the Amazon SageMaker AI Developer Guide.

If you want to create Forecast models, we recommend that you use the AmazonForecastFullAccess managed policy. If you want to use a more restrictive policy, add the following policy to your IAM role.

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": [
                "forecast:CreateAutoPredictor",
                "forecast:CreateDataset",
                "forecast:CreateDatasetGroup",
                "forecast:CreateDatasetImportJob",
                "forecast:CreateForecast",
                "forecast:CreateForecastExportJob",
                "forecast:DeleteResourceTree",
                "forecast:DescribeAutoPredictor",
                "forecast:DescribeDataset",
                "forecast:DescribeDatasetGroup",
                "forecast:DescribeDatasetImportJob",
                "forecast:DescribeForecast",
                "forecast:DescribeForecastExportJob",
                "forecast:StopResource",
                "forecast:TagResource",
                "forecast:UpdateDatasetGroup"
             ],
             "Resource": "*"
         }
    ]
}

If you want to create Amazon Bedrock models, we recommend that you use the AmazonBedrockFullAccess managed policy. If you want to use a more restrictive policy, add the following policy to your IAM role.

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Action": "bedrock:InvokeModel",
            "Resource": [
                "*",
                "arn:aws:bedrock:>region<::foundation-model/*"
            ]
        }
    ]
}
        
      

For more information about Amazon Redshift ML, see Using machine learning in Amazon Redshift, CREATE MODEL, or CREATE EXTERNAL MODEL.

Permissions for streaming ingestion

Streaming ingestion works with two services. These are Kinesis Data Streams and Amazon MSK.

Permissions required to use streaming ingestion with Kinesis Data Streams

A procedure with a managed-policy example is available at Getting started with streaming ingestion from Amazon Kinesis Data Streams.

Permissions required to use streaming ingestion with Amazon MSK

A procedure with a managed-policy example is available at Getting started with streaming ingestion from Amazon Managed Streaming for Apache Kafka.

Permissions required to use the data sharing API operations

To control access to the data sharing API operations, use IAM action-based policies. For information about how to manage IAM policies, see Managing IAM policies in the IAM User Guide.

In particular, suppose that a producer cluster administrator needs to use the AuthorizeDataShare call to authorize egress for a datashare outside of an AWS account. In this case, you set up an IAM action-based policy to grant this permission. Use the DeauthorizeDataShare call to revoke egress.

When using IAM action-based policies, you can also specify an IAM resource in the policy, such as DataShareARN. The following shows the format and an example for DataShareARN.

arn:aws:redshift:region:account-id:datashare:namespace-guid/datashare-name
arn:aws:redshift:us-east-1:555555555555:datashare:86b5169f-01dc-4a6f-9fbb-e2e24359e9a8/SalesShare

You can restrict AuthorizeDataShare access to a specific datashare by specifying the datashare name in the IAM policy.

{
  "Statement": [
    {
      "Action": [
        "redshift:AuthorizeDataShare",
      ],
      "Resource": [
        "arn:aws:redshift:us-east-1:555555555555:datashare:86b5169f-01dc-4a6f-9fbb-e2e24359e9a8/SalesShare"
      ],
      "Effect": "Deny"
    }
  ]
}

You can also restrict the IAM policy to all datashares owned by a specific producer cluster. To do this, replace the datashare-name value in the policy with a wildcard or an asterisk. Keep the cluster's namespace-guid value.

arn:aws:redshift:us-east-1:555555555555:datashare:86b5169f-01dc-4a6f-9fbb-e2e24359e9a8/*

Following is an IAM policy that prevents an entity from calling AuthorizeDataShare on the datashares owned by a specific producer cluster.

{
  "Statement": [
    {
      "Action": [
        "redshift:AuthorizeDataShare",
      ],
      "Resource": [
        "arn:aws:redshift:us-east-1:555555555555:datashare:86b5169f-01dc-4a6f-9fbb-e2e24359e9a8/*"
      ],
      "Effect": "Deny"
    }
  ]
}

DataShareARN restricts the access based on both the datashare name and the globally unique ID (GUID) for the owning cluster's namespace. It does this by specifying the name as an asterisk.

Resource policies for GetClusterCredentials

To connect to a cluster database using a JDBC or ODBC connection with IAM database credentials, or to programmatically call the GetClusterCredentials action, you need a minimum set of permissions. At a minimum, you need permission to call the redshift:GetClusterCredentials action with access to a dbuser resource.

If you use a JDBC or ODBC connection, instead of server and port you can specify cluster_id and region, but to do so your policy must permit the redshift:DescribeClusters action with access to the cluster resource.

If you call GetClusterCredentials with the optional parameters Autocreate, DbGroups, and DbName, make sure to also allow the actions and permit access to the resources listed in the following table.

For more information about resources, see Amazon Redshift resources and operations.

You can also include the following conditions in your policy:

For more information about conditions, see Specifying conditions in a policy.

Customer managed policy examples

In this section, you can find example user policies that grant permissions for various Amazon Redshift actions. These policies work when you are using the Amazon Redshift API, AWS SDKs, or the AWS CLI.

Example 1: Allow user full access to all Amazon Redshift actions and resources

The following policy allows access to all Amazon Redshift actions on all resources.

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Sid":"AllowRedshift",
      "Action": [
        "redshift:*"
      ],
      "Effect": "Allow",
      "Resource": "*"
    }
  ]
}

The value redshift:* in the Action element indicates all of the actions in Amazon Redshift.

Example 2: Deny a user access to a set of Amazon Redshift actions

By default, all permissions are denied. However, sometimes you need to explicitly deny access to a specific action or set of actions. The following policy allows access to all the Amazon Redshift actions and explicitly denies access to any Amazon Redshift action where the name starts with Delete. This policy applies to all Amazon Redshift resources in us-west-2.

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Sid":"AllowUSWest2Region",
      "Action": [
        "redshift:*"
      ],
      "Effect": "Allow",
      "Resource": "arn:aws:redshift:us-west-2:*"
    },
   {
     "Sid":"DenyDeleteUSWest2Region",
     "Action": [
        "redshift:Delete*"
      ],
      "Effect": "Deny",
      "Resource": "arn:aws:redshift:us-west-2:*"
   }
  ]
}

Example 3: Allow a user to manage clusters

The following policy allows a user to create, delete, modify, and reboot all clusters, and then denies permission to delete any clusters where the cluster name starts with protected.

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Sid":"AllowClusterManagement",
      "Action": [
        "redshift:CreateCluster",
        "redshift:DeleteCluster",
        "redshift:ModifyCluster",
        "redshift:RebootCluster"
      ],
      "Resource": [
        "*"
      ],
      "Effect": "Allow"
    },
    {
      "Sid":"DenyDeleteProtected",
      "Action": [
        "redshift:DeleteCluster"
      ],
      "Resource": [
        "arn:aws:redshift:us-west-2:123456789012:cluster:protected*"
      ],
      "Effect": "Deny"
    }
  ]
}

Example 4: Allow a user to authorize and revoke snapshot access

The following policy allows a user, for example User A, to do the following:

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Sid":"AllowSharedSnapshots",
      "Action": [
        "redshift:AuthorizeSnapshotAccess"
      ],
      "Resource": [
        "arn:aws:redshift:us-west-2:123456789012:shared/*"
      ],
      "Effect": "Allow"
    },
    {
      "Sid":"AllowRevokableSnapshot",
      "Action": [
        "redshift:RevokeSnapshotAccess"
      ],
      "Resource": [
        "arn:aws:redshift:us-west-2:123456789012:snapshot:*/revokable*"
      ],
      "Effect": "Allow"
    }
  ]
}

If User A has allowed User B to access a snapshot, User B must have a policy such as the following to allow User B to restore a cluster from the snapshot. The following policy allows User B to describe and restore from snapshots, and to create clusters. The name of these clusters must start with from-other-account.

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Sid":"AllowDescribeSnapshots",
      "Action": [
        "redshift:DescribeClusterSnapshots"
      ],
      "Resource": [
        "*"
      ],
      "Effect": "Allow"
    },
    {
      "Sid":"AllowUserRestoreFromSnapshot",
      "Action": [
        "redshift:RestoreFromClusterSnapshot"
      ],
      "Resource": [
        "arn:aws:redshift:us-west-2:123456789012:snapshot:*/*",
        "arn:aws:redshift:us-west-2:444455556666:cluster:from-other-account*"
      ],
      "Effect": "Allow"
    }
  ]
}

Example 5: Allow a user to copy a cluster snapshot and restore a cluster from a snapshot

The following policy allows a user to copy any snapshot created from the cluster named big-cluster-1, and restore any snapshot where the snapshot name starts with snapshot-for-restore.

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Sid":"AllowCopyClusterSnapshot",
      "Action": [
        "redshift:CopyClusterSnapshot"
      ],
      "Resource": [
        "arn:aws:redshift:us-west-2:123456789012:snapshot:big-cluster-1/*"
      ],
      "Effect": "Allow"
    },
    {
      "Sid":"AllowRestoreFromClusterSnapshot",
      "Action": [
        "redshift:RestoreFromClusterSnapshot"
      ],
      "Resource": [
        "arn:aws:redshift:us-west-2:123456789012:snapshot:*/snapshot-for-restore*",
        "arn:aws:redshift:us-west-2:123456789012:cluster:*"
      ],
      "Effect": "Allow"
    }
  ]
}

Example 6: Allow a user access to Amazon Redshift, and common actions and resources for related AWS services

The following example policy allows access to all actions and resources for Amazon Redshift, Amazon Simple Notification Service (Amazon SNS), and Amazon CloudWatch. It also allows specified actions on all related Amazon EC2 resources under the account.

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Sid":"AllowRedshift",
      "Effect": "Allow",
      "Action": [
        "redshift:*"
      ],
      "Resource": [
        "*"
      ]
    },
    {
      "Sid":"AllowSNS",
      "Effect": "Allow",
        "Action": [
          "sns:*"
        ],
        "Resource": [
          "*"
        ]
      },
    {
      "Sid":"AllowCloudWatch",
      "Effect": "Allow",
      "Action": [
        "cloudwatch:*"
      ],
      "Resource": [
        "*"
      ]
    },
    {
      "Sid":"AllowEC2Actions",
      "Effect": "Allow",
      "Action": [
        "ec2:AllocateAddress",
        "ec2:AssociateAddress",
        "ec2:AttachNetworkInterface",
        "ec2:DescribeAccountAttributes",
        "ec2:DescribeAddresses",
        "ec2:DescribeAvailabilityZones",
        "ec2:DescribeInternetGateways",
        "ec2:DescribeSecurityGroups",
        "ec2:DescribeSubnets",
        "ec2:DescribeVpcs"
      ],
      "Resource": [
        "*"
      ]
    }
  ]
}

Example 7: Allow a user to tag resources with the Amazon Redshift console

The following example policy allows a user to tag resources with the Amazon Redshift console using the AWS Resource Groups. This policy can be attached to a user role that invokes the new or original Amazon Redshift console. For more information about tagging, see Tag resources in Amazon Redshift.

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid":"Tagging permissions",
            "Effect": "Allow",
            "Action": [
                "redshift:DeleteTags",
                "redshift:CreateTags",
                "redshift:DescribeTags",
                "tag:UntagResources",
                "tag:TagResources"
            ],
            "Resource": "*"
        }
    ]
}

Example policy for using GetClusterCredentials

The following policy uses these sample parameter values:

The following policy enables the GetCredentials, CreateClusterUser, and JoinGroup actions. The policy uses condition keys to allow the GetClusterCredentials and CreateClusterUser actions only when the AWS user ID matches "AIDIODR4TAW7CSEXAMPLE:${redshift:DbUser}@yourdomain.com". IAM access is requested for the "testdb" database only. The policy also allows users to join a group named "common_group".

{
"Version": "2012-10-17",
  "Statement": [
    {
     "Sid":"GetClusterCredsStatement",
      "Effect": "Allow",
      "Action": [
        "redshift:GetClusterCredentials"
      ],
      "Resource": [
        "arn:aws:redshift:us-west-2:123456789012:dbuser:examplecluster/${redshift:DbUser}",
        "arn:aws:redshift:us-west-2:123456789012:dbname:examplecluster/testdb",
        "arn:aws:redshift:us-west-2:123456789012:dbgroup:examplecluster/common_group"
      ],
        "Condition": {
           "StringEquals": {
           "aws:userid":"AIDIODR4TAW7CSEXAMPLE:${redshift:DbUser}@yourdomain.com"
           }
        }
    },
    {
      "Sid":"CreateClusterUserStatement",
      "Effect": "Allow",
      "Action": [
        "redshift:CreateClusterUser"
      ],
      "Resource": [
        "arn:aws:redshift:us-west-2:123456789012:dbuser:examplecluster/${redshift:DbUser}"
      ],
      "Condition": {
        "StringEquals": {
          "aws:userid":"AIDIODR4TAW7CSEXAMPLE:${redshift:DbUser}@yourdomain.com"
        }
      }
    },
    {
      "Sid":"RedshiftJoinGroupStatement",
      "Effect": "Allow",
      "Action": [
        "redshift:JoinGroup"
      ],
      "Resource": [
        "arn:aws:redshift:us-west-2:123456789012:dbgroup:examplecluster/common_group"
      ]
    }
  ]
}