This page describes how to configure change data capture (CDC) to
stream data from an Amazon Aurora MySQL database to a [supported destination](https://docs.cloud.google.com/datastream/docs/destinations),
such as BigQuery or Cloud Storage.

> [!NOTE]
> Before configuring CDC for your Amazon Aurora MySQL database, make sure that you have an existing Aurora cluster and that you have connected to the Writer endpoint of your primary database server.

## Create a parameter group

1. Launch your Amazon RDS Dashboard.

2. In the **Navigation Drawer** , click **Parameter Groups**.

3. Click **Create Parameter Group** . The **Create Parameter Group** page appears.

4. Use the following table to populate the fields of this page:

   | Field | Description |
   |---|---|
   | Parameter group family | Select the family that matches your database. |
   | Type | Select **DB Cluster Parameter Group**. |
   | Group name | Provide a name for the parameter group. |
   | Description | Provide a description for the parameter group. |

5. Click **Create**.

6. Select the checkbox to the left of your newly created parameter group.

7. Under **Parameter group actions** , click **Edit**.

8. Set the following parameter values:

   - For **GTID-based** replication:

     | Parameter | Value |
     |---|---|
     | binlog_format | `ROW` |
     | log_bin_use_v1_row_events | `1` |
     | read_only | `0` |
     | net_read_timeout | `3600` |
     | net_write_timeout | `3600` |
     | wait_timeout | `86400` |
     | expire-logs-days | `7` |
     | binlog_row_image | `FULL` |
     | max_allowed_packet | `1G` (recommended value) |
     | log-replica-updates | `1` |
     | enforce-gtid-consistency | `ON` |
     | gtid_mode | `ON` You can only set this parameter to `ON` after you set `enforce-gtid-consistency` to `ON`. In general, `gtid_mode` can have the following values: - `OFF` - `OFF_PERMISSIVE` - `ON_PERMISSIVE` - `ON` You can only change the values of the `gtid_mode` parameter one step at a time. For example, if the current mode is set to `OFF`, do the following: 1. Change the setting to `OFF_PERMISSIVE`. 2. Change the setting to `ON_PERMISSIVE`. 3. Change the setting to `ON`. For more information, see [MySQL documentation](https://dev.mysql.com/doc/refman/8.4/en/replication-options-gtids.html#sysvar_gtid_mode). |

   - For **binlog-based** replication:

     | Parameter | Value |
     |---|---|
     | binlog_format | `ROW` |

9. Click **Save Changes**.

## Assign the parameter group to the database instance

1. Launch your Amazon RDS Dashboard.

2. In the **Navigation Drawer** , click **Databases**, and then select your database instance.

3. From the **Instance Actions** menu, select **Modify** . The **Modify DB Instance** dialog appears.

4. In the **Additional configuration** section, select the database cluster parameter group that you [created](https://docs.cloud.google.com/datastream/docs/configure-amazon-aurora-mysql#create-a-parameter-group-aurora).

5. Set the **Backup retention period** to **7 days**.

6. Click **Continue**.

7. In the **Scheduling of Modifications** pane, select the **Apply immediately** option.

## Set the binary log retention period

1. Enter the following MySQL command:

   ```sql
   call mysql.rds_set_configuration('binlog retention hours', 168);
   ```

   > [!NOTE]
   > You're entering this command because you want to ensure proper replication by configuring your system to retain binary logs for a minimum of 7 days (or 168 hours).

2. Restart your MySQL server so that the changes you made can take effect.

## Create a Datastream user

1. Create a Datastream user:

   ```sql
   CREATE USER 'datastream'@'%' IDENTIFIED BY 'YOUR_PASSWORD';
   ```
2. Grant the following privileges to them:

   ```sql
   GRANT REPLICATION SLAVE, SELECT, REPLICATION CLIENT ON *.* TO 'datastream'@'%';
   GRANT EXECUTE ON PROCEDURE mysql.rds_show_configuration TO 'datastream'@'%';
   FLUSH PRIVILEGES;
   ```

## What's next

- Learn more about how Datastream works with [MySQL sources](https://docs.cloud.google.com/datastream/docs/sources-mysql).