Cloud Data Ingestion for Amazon Redshift

This page covers everything you need to sync data from Amazon Redshift to Braze using Cloud Data Ingestion (CDI): the resources to create in Redshift, how to connect the source and create a sync in Braze, the source table for each data type, and Redshift-specific troubleshooting.

Setup overview

1. In your Redshift instance, set up the tables or views you want to sync to Braze.
2. Create a Redshift user for Braze with one of the supported authentication methods, and grant it access to your tables.
3. Make sure Braze access is allowed to the Redshift tables you want to sync. Braze connects to Redshift over the internet.
4. Create a new source and sync in the Braze dashboard.
5. Validate the integration and start the sync.

note

Rows processed per sync depend on your warehouse performance, network latency, and how much new data matches the sync query. Use the integration Sync history in the dashboard to see duration and row counts for recent runs.

Step 1: Set up your source table

The steps in this section create a table with a PAYLOAD column. If you plan to map columns with Visual or write a query with SQL on the Data definition step, you can use an existing table or view instead. Skip the table creation step, and grant Braze read access to the tables or views you want to sync. For a comparison of options, see Choose a data definition option.

You can name the database, schema, and table as you’d like, but the column names should match the definitions in the following tabs. You may also use a view or a materialized view instead of a table.

Optionally, set up a new database and schema to hold your source table:

1
2

CREATE DATABASE BRAZE_CLOUD_PRODUCTION;
CREATE SCHEMA BRAZE_CLOUD_PRODUCTION.INGESTION;

note

If you use a PAYLOAD column, your source table or view can include columns that aren’t listed in the following tabs (for example, auditing or hashing). Braze reads only the columns described in these tabs; other columns are not used during Cloud Data Ingestion syncs.

Select the tab for the data type you want to sync.

• user attributes
• custom events
• purchase events
• catalog items
• canvas triggers
• user deletions

User attributes, custom events, and purchase events all use the same source table shape. Only the contents of PAYLOAD differ.

1
2
3
4
5
6
7
8
9
10
11
12
13
14

CREATE TABLE BRAZE_CLOUD_PRODUCTION.INGESTION.USERS_ATTRIBUTES_SYNC (
   updated_at timestamptz default sysdate,
   --at least one of external_id, alias_name and alias_label, or braze_id is required
   external_id varchar,
   --if using user alias, both alias_name and alias_label are required
   alias_label varchar,
   alias_name varchar,
   --braze_id can only be used to update existing users created through the Braze SDK
   braze_id varchar,
   --If you include both email and phone, email is used as the primary identifier
   email varchar,
   phone varchar,
   payload varchar(max)
)

• UPDATED_AT - The time this row was updated in or added to the table. Braze syncs rows where UPDATED_AT is later than the last synced value. Rows at the exact boundary timestamp may be re-synced if new rows share that same timestamp.
• User identifier columns - Your table may contain one or more user identifier columns. Each row should only contain one identifier (either external_id, the combination of alias_name and alias_label, braze_id, email or phone). A source table may have columns for one, two, three, four, or all five identifier types.
  • EXTERNAL_ID - This identifies the user you want to update. This should match the external_id value used in Braze.
  • ALIAS_NAME and ALIAS_LABEL - These two columns create a user alias object. alias_name should be a unique identifier, and alias_label specifies the type of alias. Users may have multiple aliases with different labels but only one alias_name per alias_label.
  • BRAZE_ID - The Braze user identifier. This is generated by the Braze SDK, and new users cannot be created using a Braze ID through Cloud Data Ingestion. To create new users, specify an external user ID or user alias.
  • EMAIL - The user’s email address. If multiple profiles with the same email address exist, the most recently updated profile is prioritized for updates. If you include both email and phone, email is used as the primary identifier.
  • PHONE - The user’s phone number. If multiple profiles with the same phone number exist, the most recently updated profile is prioritized for updates.
• PAYLOAD - This is a JSON string of the fields you want to sync to the user in Braze.

For attribute syncs, PAYLOAD follows the user attributes object format. You can include nested custom attributes, arrays of objects, and subscription group statuses. For payload examples, see Table setup.

Custom events use the same source table as user attributes. Create the table shown in the User attributes tab, then populate PAYLOAD with an events object.

To sync events, an event name is required. Format the time field as an ISO 8601 string or in yyyy-MM-dd'T'HH:mm:ss:SSSZ format. If the time field is not present, Braze uses the UPDATED_AT column value as the event time. Other fields, including app_id and properties, are optional.

You can sync one event per row.

1
2
3
4
5
6
7
8
9

{
    "app_id" : "your-app-id",
    "name" : "rented_movie",
    "time" : "2013-07-16T19:20:45+01:00",
    "properties": {
        "movie": "The Sad Egg",
        "director": "Alex Smith"
    }
}

Purchase events use the same source table as user attributes. Create the table shown in the User attributes tab, then populate PAYLOAD with a purchases object.

To sync purchase events, product_id, currency, and price are required. Format the optional time field as an ISO 8601 string or in yyyy-MM-dd'T'HH:mm:ss:SSSZ format. If the time field is not present, Braze uses the UPDATED_AT column value as the event time. Other fields, including app_id, quantity, and properties, are optional.

You can sync one purchase event per row.

1
2
3
4
5
6
7
8
9
10
11
12

{
    "app_id" : "11ae5b4b-2445-4440-a04f-bf537764c9ad",
    "product_id" : "Completed Order",
    "currency" : "USD",
    "price" : 219.98,
    "time" : "2013-07-16T19:20:30+01:00",
    "properties" : {
        "products" : [ { "name": "Monitor", "category": "Gaming", "product_amount": 19.99 },
        { "name": "Gaming Keyboard", "category": "Gaming ", "product_amount": 199.99 }
        ]
    }
}

Before you create a catalog sync, create the target catalog in Braze. For the full catalog workflow, see Sync and delete catalog data.

1
2
3
4
5
6
7
8
9

CREATE TABLE BRAZE_CLOUD_PRODUCTION.INGESTION.CATALOGS_SYNC (
   updated_at timestamptz default sysdate not null,
   --ID of the catalog item to be created or updated
   id varchar not null,
   --Catalog fields and values that should be added or updated
   payload varchar(max),
   --The catalog item associated with this ID should be deleted
   deleted boolean
)

Grant the Braze user access to this table alongside your other source tables:

1

GRANT SELECT ON TABLE CATALOGS_SYNC TO braze_user;

The name of the integration must exactly match the name of the catalog you created in Braze.

Braze recommends creating a view from your catalog data so the source table fully refreshes each time a sync runs. For example, if you have a table of product data (product_catalog_1) with product_id and three additional attributes, you could sync the following:

1
2
3
4
5
6
7
8
9
10
11
12
13

CREATE TABLE BRAZE_CLOUD_PRODUCTION.INGESTION.CATALOGS_SYNC AS
SELECT
    CURRENT_TIMESTAMP as UPDATED_AT,
    Product_id as id,
    JSON_SERIALIZE(
        OBJECT (
            'attribute_1',
            attribute_1,
            'attribute_2',
            attribute_2,
            'attribute_3',
            attribute_3)
    ) as PAYLOAD FROM "product_catalog_1";

Canvas triggers power zero-copy personalization. For the destination Canvas setup and considerations, see Zero-copy personalization.

1
2
3
4
5
6
7
8
9

CREATE TABLE BRAZE_CLOUD_PRODUCTION.INGESTION.CANVAS_TRIGGERS_SYNC (
    updated_at timestamptz default sysdate not null,
    --at least one of external_id or alias_name and alias_label is required
    external_id varchar not null,
    --if using user alias, both alias_name and alias_label are required
    alias_label varchar,
    alias_name varchar,
    properties varchar(max)
 );

Canvas trigger syncs differ from other user data syncs in three ways:

• Only external ID or user alias identifiers are supported. Email and phone numbers are not supported identifiers.
• Only existing Braze users can be synced. New users cannot be created.
• PROPERTIES replaces the PAYLOAD column. This is a JSON string of the fields you want to use as Canvas entry properties for personalization.

Your Canvas triggers table needs the following columns:

• UPDATED_AT - The time this row was updated or added to the table. Braze syncs rows where UPDATED_AT is later than the last synced value. Rows at the exact boundary timestamp may be re-synced if new rows share that same timestamp.
• Either EXTERNAL_ID or ALIAS_NAME and ALIAS_LABEL as the user identifier column. These identify the users for whom you want to trigger Canvas messaging.
  • EXTERNAL_ID - Identifies the user to enter into the Canvas. This should match the external_id value used in Braze.
  • ALIAS_NAME and ALIAS_LABEL - These columns create a user alias object. alias_name should be a unique identifier, and alias_label specifies the alias type. Users may have multiple aliases with different labels, but only one alias_name per alias_label.
• PROPERTIES - A JSON string of fields to make available as personalization properties in your Canvas. This should contain user-specific information.

note

Properties are not required for every row or user. However, properties values must be a valid JSON string. Input an empty {} string if there are no properties for the row.

Grant the Braze user access to this table:

1

GRANT SELECT ON TABLE CANVAS_TRIGGERS_SYNC TO braze_user;

To get started, select the Canvas Triggers data type when creating a new sync.

Source tables for user deletes should include one or more user identifier types and an UPDATED_AT timestamp. Payload columns aren’t supported for user delete data.

• UPDATED_AT - The time this row was updated or added to the table. Braze syncs rows where UPDATED_AT is later than the last synced value. Rows at the exact boundary timestamp may be re-synced if new rows share that same timestamp.
• User identifier columns - Each row should only contain one identifier: either external_id, the combination of alias_name and alias_label, or braze_id. A source table may contain columns for one, two, or all three identifier types.
  • EXTERNAL_ID - This identifies the user you want to update. This should match the external_id value used in Braze.
  • ALIAS_NAME and ALIAS_LABEL - These two columns create a user alias object. alias_name should be a unique identifier, and alias_label specifies the type of alias. Users may have multiple aliases with different labels but only one alias_name per alias_label.
  • BRAZE_ID - The Braze user identifier. This is generated by the Braze SDK and new users cannot be created using a Braze ID through Cloud Data Ingestion. To create new users, specify an external user ID or user alias.

important

Do not include a PAYLOAD column in your table for user removal. To prevent accidental, permanent removal of users, a sync fails if a payload column is provided in the source table. Any other columns are allowed but are ignored by Braze.

1
2
3
4
5
6
7
8
9
10

CREATE TABLE BRAZE_CLOUD_PRODUCTION.INGESTION.USERS_DELETES (
   updated_at timestamptz default sysdate,
   --at least one of external_id, alias_name and alias_label, or braze_id is required
   external_id varchar,
   --if using user alias, both alias_name and alias_label are required
   alias_label varchar,
   alias_name varchar,
   --braze_id can only be used to update existing users created through the Braze SDK
   braze_id varchar
);

On the Create import sync page, set the Data Type to Delete Users so that the proper actions are taken during the integration run. For more information, see Delete users with Cloud Data Ingestion.

warning

Deleting user profiles cannot be undone. The delete action permanently removes users, which may cause discrepancies in your data. For details, see Effects of deleting user profiles.

Step 2: Set up authentication

Braze connects to Redshift as a database user. Create the user with one of the following authentication methods:

Username and password

Braze connects to Redshift as a database user with a password. Create a user for Braze:

1

CREATE USER braze_user PASSWORD '{password}';

Keep the username and password in a safe place until you need to enter them in the Braze dashboard when you create the source.

Step 3: Grant permissions to the user

Grant the user you created for Braze access to your source table:

1
2

GRANT USAGE ON SCHEMA BRAZE_CLOUD_PRODUCTION.INGESTION to braze_user;
GRANT SELECT ON TABLE USERS_ATTRIBUTES_SYNC TO braze_user;

These are the minimum required permissions for this user. If creating multiple CDI integrations, you may want to grant permissions to a schema or manage permissions using a group.

Step 4: Allow access to Braze IPs

If you have a firewall or other network policies, you must give Braze network access to your Redshift instance. An example of the Redshift URL endpoint is “example-cluster.ap-northeast-2.redshift.amazonaws.com”.

Some important things to know:

• You may also need to change your security groups to allow Braze to access your data in Redshift.
• Make sure to explicitly allow inbound traffic on the IPs in the table and on the port used to query your Redshift cluster (default is 5439). You should explicitly allow Redshift TCP connectivity on this port even if the inbound rules are set to “allow all”.
• The endpoint for the Redshift cluster must be publicly accessible for Braze to connect to your cluster.
  • If you don’t want your Redshift cluster to be publicly accessible, you can set up a VPC and EC2 instance to use an SSH tunnel to access the Redshift data. For more information, see the AWS Knowledge Center post.

Allow access from the following IPs corresponding to your Braze dashboard’s region.

• united states (us)
• european union (eu)
• australia (au)
• indonesia (id)
• japan (jp)
• south korea (kr)

Step 5: Create a new source in the Braze dashboard

In the Braze dashboard, go to Data Settings > Cloud Data Ingestion > Sources, select Add data source, and then select Amazon Redshift.

Step 5.1: Add Redshift connection information and source table

Choose a name for your source and input your Redshift credentials and configuration. Select the authentication method you set up in Step 2, and enter its credentials:

• Username and password: Username and Password

If you’re using a private network tunnel, toggle the slider and input the tunnel information. Then, proceed to the next step.

note

In the Braze dashboard, the Database name field only accepts letters (A–Z, a–z), numbers (0–9), and underscores (_), even though Amazon Redshift supports additional characters in database identifiers.

Step 5.2: Test connection and connect to source

Next, select Test connection. Once successful, finalize remaining settings and click Connect to Source. If the connection fails, an error message appears to help troubleshoot the issue.

Step 6: Create a new sync in the Braze dashboard

Go to Data Settings > Cloud Data Ingestion > Syncs, and select Create data sync.

Configure sync details and define your data

Enter a name for your sync. Then, select an active source and a sync data type. Next, select Visual or Table and enter your source table, or select SQL to write a custom SQL query. Finally, select Preview and validate.

After validation succeeds, a preview of the data appears. Select Next: Notifications to continue. If validation fails, an error message appears to help troubleshoot the issue.

note

You must successfully validate a sync before progressing to next steps. If you need to close out of the sync creation page, select Save as draft to keep your work in progress.

Add notification preferences

Input contact email(s) for sync error notifications. Braze uses this contact information to send notifications about integration errors, such as unexpected loss of table access.

Contact emails only receive notifications of global or sync-level errors such as missing tables, permissions, and others. They do not receive row-level issues. Global errors indicate critical problems with the connection that prevent syncs from running.

Such problems can include the following:

• Connectivity issues
• Lack of resources
• Permissions issues
• Catalog tier is out of space (catalog syncs only)

Set the schedule

Lastly, configure your sync as non-recurring or recurring.

Non-recurring syncs run only when you select Sync Now in the dashboard or call the Trigger a sync endpoint.

Recurring syncs can run as often as every 5 minutes or as rarely as once per month. By default, the shortest interval you can set in the dashboard is 15 minutes, which helps manage your warehouse compute costs and request volume. To sync as often as every 5 minutes, contact Braze Support or your customer success manager. Braze schedules recurring syncs in UTC.

note

You must successfully validate an integration before it can move from Draft to Active state. If you close the creation page, your integration is saved, and you can revisit the details page to make changes and validate.

Run the sync

When activated, your sync runs on the schedule configured during setup. If you want to run the sync outside its schedule or to fetch the most recent data, select Sync Now. This run does not impact regularly scheduled future syncs. To start a sync automatically when your data updates, call the Trigger a sync endpoint from your pipeline.

Set up additional integrations or users (optional)

You can set up multiple integrations with Braze. When creating additional syncs, you can reuse existing credentials if connecting to the same Redshift account.

If you reuse the same user across integrations, you cannot delete the user in the Braze dashboard until it’s removed from all active syncs.

Use Redshift as a connected source

A connected source is a zero-copy alternative to syncing data into Braze. Braze queries your Redshift instance directly to build CDI segments, without copying the underlying data. For concepts, limitations, and cost implications, see Connected sources.

Step 1: Create the user and grant permissions

Create a user for Braze with one of the methods in Step 2: Set up authentication. Then, grant it the following permissions:

1
2
3

GRANT USAGE ON SCHEMA BRAZE_CLOUD_PRODUCTION.INGESTION to braze_user;
GRANT CREATE ON SCHEMA BRAZE_CLOUD_PRODUCTION.INGESTION to braze_user;
GRANT SELECT ON TABLE USERS_ATTRIBUTES_SYNC TO braze_user;

Create a user for your connected source to use. This user is used to generate the list of tables available in your CDI Segment Extensions, and to query source tables to create new segments. After the connected source is created, Braze discovers the names and description of all tables available to the user in the source schema. If creating multiple CDI integrations, you may want to grant permissions to a schema or manage permissions using a group.

You may choose to grant access to all tables in a schema, or grant privileges only to specific tables. Whichever tables the Braze role has access to are available to query in the CDI Segment Extension. Be sure to grant access to any new tables to the user when they’re created, or set default permissions for the user.

The create table permission is required so Braze can create a table with your CDI Segment Extension query results before updating the segment in Braze. Braze creates a temporary table per segment, which only persists while Braze updates the segment.

Network access setup matches Step 4.

Step 2: Create the connected source in Braze

Go to Data Settings > Cloud Data Ingestion > Connected Sources, select Add data source, and then select Amazon Redshift.

In Setup source, enter the following:

• Credentials: Redshift Host URL, Port, and the fields for your authentication method. Enter Username and Password.

• Configuration: Database and Schema

If needed, enable Connect with SSH Tunnel and enter Tunnel Host, Tunnel Port, and Tunnel Username.

Choose a name for the connected source. This name is used in the list of available sources when you create a new CDI Segment Extension.

Configure a maximum runtime for this source. Braze automatically aborts any queries that exceed the maximum runtime. The maximum runtime allowed is 60 minutes; a lower runtime reduces costs incurred on your Redshift account. This setting applies to queries run through this source, including syncs and CDI Segment Extensions that use it.

Step 3: Finalize the Redshift configuration

If your credentials have Connect with SSH Tunnel selected, take note of the RSA public key in the Test connection step and add it to the SSH tunnel user.

After you add the key to the user, select Test Connection in Braze, and then select Done. Your connected source is now created and ready to use in CDI Segment Extensions.

You may set up multiple connected sources with Braze, but each source should be configured to connect a different schema. When creating additional sources, you may reuse existing credentials if connecting to the same Redshift account.

Format a payload from existing columns

If you prefer to store each attribute in its own column internally, you need to convert those columns to a JSON string to populate the sync with Braze. Use this query in Redshift to format source columns into CDI fields.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

CREATE TABLE "EXAMPLE_USER_DATA"
    (attribute_1 string,
     attribute_2 string,
     attribute_3 number,
     my_user_id string);

SELECT
    CURRENT_TIMESTAMP as UPDATED_AT,
    my_user_id as EXTERNAL_ID,
    JSON_SERIALIZE(
        OBJECT (
            'attribute_1',
            attribute_1,
            'attribute_2',
            attribute_2,
            'yet_another_attribute',
            attribute_3)
    ) as PAYLOAD FROM "EXAMPLE_USER_DATA";

For more optimization guidance, see Cloud Data Ingestion best practices.

Troubleshooting

Test Connection runs slow

Test Connection is running on your data warehouse, so increasing warehouse capacity may improve its speed. Using a serverless SQL instance minimizes warmup time and improves query throughput, but may result in slightly higher integration costs.

Permission denied for relation {table_name}

If you receive this error:

• Grant the usage permission on the schema for that user.
• Grant the select permission on the table for that user.

Create Connection Error

If you receive this error, verify the Redshift endpoint and port are correct.

Create SSH Tunnel Error

If you receive this error:

• Verify the public key on your Braze dashboard is on the EC2 host used for SSH tunneling.
• Verify your username is correct.
• Verify the SSH tunnel is correct.

Invalid snapshot identifier

If Braze returns an Invalid snapshot identifier error during Test connection or sync setup, Redshift can’t resolve the snapshot reference used when your source object is queried.

In Redshift, a snapshot is a point-in-time backup of a cluster. Each snapshot has a unique identifier used by Redshift to reference that backup state. For more information, see Amazon Redshift snapshots and backups.

This error can occur when metadata changes while Braze validates the source object, such as during snapshot copy, restore, or replication-related operations. For more information, see copying snapshots to another AWS Region and restoring a cluster from a snapshot.

To troubleshoot:

1. Verify source settings in Braze, including cluster endpoint, database, schema, and object name.
2. Run the same query directly in Redshift to confirm the table or view is readable and stable.
3. Retry after active snapshot, restore, resize, or replication activity finishes.
4. If the issue persists, query a materialized view instead of a frequently changing base table.

A materialized view stores precomputed query results that you can refresh on a schedule, which can make reads more stable for CDI syncs. For more information, see materialized views in Amazon Redshift.

Example:

1
2
3
4
5

CREATE MATERIALIZED VIEW ingestion.users_attributes_mv AS
SELECT updated_at, external_id, alias_label, alias_name, braze_id, email, phone, payload
FROM ingestion.users_attributes_sync;

REFRESH MATERIALIZED VIEW ingestion.users_attributes_mv;

After you create the materialized view, use the materialized view name as the source object in your Braze CDI sync instead of the base table.

Where to run SQL checks

Run validation checks directly in the Redshift SQL editor, against the same table or view used by your CDI integration. Use Query Editor v2. For more information, see Using Amazon Redshift Query Editor v2.

Related pages

• Table setup for required columns and payload formatting
• Best practices for sync scheduling and data point efficiency
• Sync logs and observability for monitoring sync runs
• FAQ for common questions across all sources