Data protection policy operations - Amazon Simple Notification Service
Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

Data protection policy operations

The following are examples of data protection policies that you can use to audit and deny sensitive data. For a complete tutorial that includes an example application, see the Introducing message data protection for Amazon SNS blog post.

Audit operation

The Audit operation samples topic inbound messages, and logs the sensitive data findings in an Amazon destination. The sample rate can be an integer between 0–99. This operation requires one of the following types of logging destinations:

  1. FindingsDestination – The logging destination when the Amazon SNS topic finds sensitive data in the payload.

  2. NoFindingsDestination – The logging destination when the Amazon SNS topic doesn't find sensitive data in the payload.

You can use the following Amazon Web Services in each of the following log destination types:

{ "Operation": { "Audit": { "SampleRate": "99", "FindingsDestination": { "CloudWatchLogs": { "LogGroup": "/aws/vendedlogs/log-group-name" }, "Firehose": { "DeliveryStream": "delivery-stream-name" }, "S3": { "Bucket": "bucket-name" } }, "NoFindingsDestination": { "CloudWatchLogs": { "LogGroup": "/aws/vendedlogs/log-group-name" }, "Firehose": { "DeliveryStream": "delivery-stream-name" }, "S3": { "Bucket": "bucket-name" } } } } }

Required permissions when specifying log destinations

When you specify logging destinations in the data protection policy, you must add the following permissions to the IAM identity policy of the IAM principal that is calling the Amazon SNS PutDataProtectionPolicy API, or the CreateTopic API with the --data-protection-policy parameter.

Audit destination IAM permission
















Extra actions are required for using Amazon S3 bucket with SSE-KMS encryption enabled.

{ "Version": "2012-10-17", "Statement": [ { "Effect": "Allow", "Action": [ "logs:CreateLogDelivery", "logs:GetLogDelivery", "logs:UpdateLogDelivery", "logs:DeleteLogDelivery", "logs:ListLogDeliveries" ], "Resource": [ "*" ] }, { "Effect": "Allow", "Action": [ "logs:PutResourcePolicy", "logs:DescribeResourcePolicies", "logs:DescribeLogGroups" ], "Resource": [ "arn:aws-cn:logs:region:account-id:SampleLogGroupName:*:*" ] }, { "Effect": "Allow", "Action": [ "iam:CreateServiceLinkedRole", "firehose:TagDeliveryStream" ], "Resource": "*" }, { "Effect": "Allow", "Action": [ "s3:PutBucketPolicy", "s3:GetBucketPolicy" ], "Resource": [ "arn:aws-cn:s3:::bucket-name" ] } ] }

Required key policy for use with SSE-KMS

If you use an Amazon S3 bucket as a log destination, you can protect the data in your bucket by enabling either Server-Side Encryption with Amazon S3-Managed Keys (SSE-S3), or Server-Side Encryption with Amazon KMS keys (SSE-KMS). For more information, see Protecting data using server-side encryption in the Amazon S3 User Guide.

If you choose SSE-S3, no additional configuration is required. Amazon S3 handles the encryption key.

If you choose SSE-KMS, you must use a customer managed key. You must update the key policy for your customer managed key so that the log delivery account can write to your S3 bucket. For more information about the required key policy for use with SSE-KMS, see Amazon S3 bucket server-side encryption in the Amazon CloudWatch Logs User Guide.

Audit destination log example

In the following example, callerPrincipal is used to identify the source of the sensitive content, and messageID is used as a reference to check against the Publish API response.

{ "messageId": "34d9b400-c6dd-5444-820d-fbeb0f1f54cf", "auditTimestamp": "2022-05-12T2:10:44Z", "callerPrincipal": "arn:aws-cn:iam::123412341234:role/Publisher", "resourceArn": "arn:aws-cn:sns:us-east-1:123412341234:PII-data-topic", "dataIdentifiers": [ { "name": "Name", "count": 1, "detections": [ { "start": 1, "end": 2 } ] }, { "name": "PhoneNumber", "count": 2, "detections": [ { "start": 3, "end": 4 }, { "start": 5, "end": 6 } ] } ] }

Audit operation metrics

When an audit operation has specified the FindingsDestination or the NoFindingsDestination property, the topic owners also receive CloudWatch MessagesWithFindings and MessagesWithNoFindings metrics.

Example of an audit displaying data over a specified period of time.

De-identify operation

The De-identify operation masks or redacts sensitive data from published or delivered messages. This operation is available for both inbound and outbound messages, and requires one of the following types of configurations:

  • MaskConfig – Mask using a supported character from the following table. For example, ssn: 123-45-6789 becomes ssn: ###########.

    { "Operation": { "Deidentify": { "MaskConfig": { "MaskWithCharacter": "#" } } }
    Supported mask character Name
    * Asterisk
    A-Z, a-z, and 0-9 Alphanumeric
    ! Exclamation mark
    $ Dollar sign
    % Percent sign
    & Ampersand


    + Plus sign
    , Comma
    - Hyphen
    . Period

    Slash, back slash

    # Number sign
    : Colon
    ; Semicolon
    =, <>

    Equals. less or greater than

    @ At sign
    [] Brackets
    ^ Caret symbol
    _ Underscore
    ` Backtick
    | Vertical bar
    ~ Tilde symbol
  • RedactConfig – Redact by removing the data entirely. For example, ssn: 123-45-6789 becomes ssn: .

    { "Operation": { "Deidentify": { "RedactConfig": {} } }

On an inbound message, the sensitive data is de-identified after the audit operation, and the SNS:Publish API caller receives the following invalid parameter error when the entire message is sensitive.

Error code: AuthorizationError ...

Deny operation

The Deny operation interrupts either the Publish API request or the delivery of the message if the message contains sensitive data. The Deny operation object is empty, as it doesn't require additional configuration.

"Operation": { "Deny": {} }

On an inbound message, the SNS:Publish API caller receives an authorization error.

Error code: AuthorizationError ...

On an outbound message, the Amazon SNS topic does not deliver the message to the subscription. To track unauthorized deliveries, enable the topic’s delivery status logging. The following is an example of a delivery status log:

{ "notification": { "messageMD5Sum": "29638742ffb68b32cf56f42a79bcf16b", "messageId": "34d9b400-c6dd-5444-820d-fbeb0f1f54cf", "topicArn": "arn:aws-cn:sns:us-east-1:123412341234:PII-data-topic", "timestamp": "2022-05-12T2:12:44Z" }, "delivery": { "deliveryId": "98236591c-56aa-51ee-a5ed-0c7d43493170", "destination": "arn:aws-cn:sqs:us-east-1:123456789012:NoNameAccess", "providerResponse": "The topic's data protection policy prohibits this message from being delivered to <subscription-arn>", "dwellTimeMs":20, "attempts":1, "statusCode": 403 }, "status": "FAILURE" }