Data filtering for Amazon RDS zero-ETL integrations

Amazon RDS zero-ETL integrations support data filtering, which lets you control which data is replicated from your source Amazon RDS database to your target data warehouse. Instead of replicating the entire database, you can apply one or more filters to selectively include or exclude specific tables. This helps you optimize storage and query performance by ensuring that only relevant data is transferred. Currently, filtering is limited to the database and table levels. Column- and row-level filtering are not supported.

Data filtering can be useful when you want to:

• Join certain tables from two or more different source databases, and you don't need complete data from either database.

• Save costs by performing analytics using only a subset of tables rather than an entire fleet of databases.

• Filter out sensitive information—such as phone numbers, addresses, or credit card details—from certain tables.

You can add data filters to a zero-ETL integration using the AWS Management Console, the AWS Command Line Interface (AWS CLI), or the Amazon RDS API.

If the integration has a provisioned cluster as its target, the cluster must be on patch 180 or higher to use data filtering.

Format of a data filter

You can define multiple filters for a single integration. Each filter either includes or excludes any existing and future database tables that match one of the patterns in the filter expression. Amazon RDS zero-ETL integrations use Maxwell filter syntax for data filtering.

Each filter has the following elements:

Element	Description
Filter type	An Include filter type includes all tables that match one of the patterns in the filter expression. An Exclude filter type excludes all tables that match one of the patterns.
Filter expression	A comma-separated list of patterns. Expressions must use Maxwell filter syntax.
Pattern	A filter pattern in the format database.table for RDS for MySQL, or database.schema.table for RDS for PostgreSQL. You can specify literal names, or define regular expressions. Note For RDS for MySQL, regular expressions are supported in both the database and table name. For RDS for PostgreSQL, regular expressions are supported only in the schema and table name, not in the database name. You can't include column-level filters or denylists. A single integration can have a maximum of 99 total patterns. In the console, you can enter patterns within a single filter expression, or spread them out among multiple expressions. A single pattern can't exceed 256 characters in length.

Important

If you select an RDS for PostgreSQL source database, you must specify at least one data filter pattern. At minimum, the pattern must include a single database (database-name.*.*) for replication to the target data warehouse.

The following image shows the structure of RDS for MySQL data filters in the console:

Important

Do not include personally identifying, confidential, or sensitive information in your filter patterns.

Data filters in the AWS CLI

When using the AWS CLI to add a data filter, the syntax differs slightly from the console. You must assign a filter type (Include or Exclude) to each pattern individually, so you can't group multiple patterns under one filter type.

For example, in the console you can group the following comma-separated patterns under a single Include statement:

RDS for MySQL

mydb.mytable, mydb./table_\d+/

RDS for PostgreSQL

mydb.myschema.mytable, mydb.myschema./table_\d+/

However, when using the AWS CLI, the same data filter must be in the following format:

RDS for MySQL

'include: mydb.mytable, include: mydb./table_\d+/'

RDS for PostgreSQL

'include: mydb.myschema.mytable, include: mydb.myschema./table_\d+/'

Filter logic

If you don't specify any data filters in your integration, Amazon RDS assumes a default filter of include:*.*, which replicates all tables to the target data warehouse. However, if you add at least one filter, the default logic switches to exclude:*.*, which excludes all tables by default. This lets you explicitly define which databases and tables to include in replication.

For example, if you define the following filter:

'include: db.table1, include: db.table2'

Amazon RDS evaluates the filter as follows:

'exclude:*.*, include: db.table1, include: db.table2'

Therefore, Amazon RDS only replicates table1 and table2 from the database named db to the target data warehouse.

Filter precedence

Amazon RDS evaluates data filters in the order you specify. In the AWS Management Console, it processes filter expressions from left to right and top to bottom. A second filter or an individual pattern that follows the first can override it.

For example, if the first filter is Include books.stephenking, it includes only the stephenking table from the books database. However, if you add a second filter, Exclude books.*, it overrides the first filter. This prevents any tables from the books index from being replicated to the target data warehouse.

When you specify at least one filter, the logic starts by assuming exclude:*.* by default, which automatically excludes all tables from replication. As a best practice, define filters from broadest to most specific. Start with one or more Include statements to specify the data to replicate, then add Exclude filters to selectively remove certain tables.

The same principle applies to filters that you define using the AWS CLI. Amazon RDS evaluates these filter patterns in the order that you specify them, so a pattern might override one that you specify before it.

RDS for MySQL examples

The following examples demonstrate how data filtering works for RDS for MySQL examples zero-ETL integrations:

• Include all databases and all tables:

  'include: *.*'

• Include all tables within the books database:

  'include: books.*'

• Exclude any tables named mystery:

  'include: *.*, exclude: *.mystery'

• Include two specific tables within the books database:

  'include: books.stephen_king, include: books.carolyn_keene'

• Include all tables in the books database, except for those containing the substring mystery:

  'include: books.*, exclude: books./.*mystery.*/'

• Include all tables in the books database, except those starting with mystery:

  'include: books.*, exclude: books./mystery.*/'

• Include all tables in the books database, except those ending with mystery:

  'include: books.*, exclude: books./.*mystery/'

• Include all tables in the books database that start with table_, except for the one named table_stephen_king. For example, table_movies or table_books would be replicated, but not table_stephen_king.

  'include: books./table_.*/, exclude: books.table_stephen_king'

RDS for PostgreSQL examples

The following examples demonstrate how data filtering works for RDS for PostgreSQL zero-ETL integrations:

• Include all tables within the books database:

  'include: books.*.*'

• Exclude any tables named mystery in the books database:

  'include: books.*.*, exclude: books.*.mystery'

• Include one table within the books database in the mystery schema, and one table within employee database in the finance schema:

  'include: books.mystery.stephen_king, include: employee.finance.benefits'

• Include all tables in the books database and science_fiction schema, except for those containing the substring king:

  'include: books.science_fiction.*, exclude: books.*./.*king.*/

• Include all tables in the books database, except those with a schema name starting with sci:

  'include: books.*.*, exclude: books./sci.*/.*'

• Include all tables in the books database, except those in the mystery schema ending with king:

  'include: books.*.*, exclude: books.mystery./.*king/'

• Include all tables in the books database that start with table_, except for the one named table_stephen_king. For example, table_movies in the fiction schema and table_books in the mystery schema are replicated, but not table_stephen_king in either schema:

  'include: books.*./table_.*/, exclude: books.*.table_stephen_king'

RDS for Oracle examples

The following examples demonstrate how data filtering works for RDS for Oracle zero-ETL integrations:

• Include all tables within the books database:

  'include: books.*.*'

• Exclude any tables named mystery in the books database:

  'include: books.*.*, exclude: books.*.mystery'

• Include one table within the books database in the mystery schema, and one table within employee database in the finance schema:

  'include: books.mystery.stephen_king, include: employee.finance.benefits'

• Include all tables in the mystery schema within the books database:

  'include: books.mystery.*'

Case sensitivity considerations

Oracle Database and Amazon Redshift handle object name casing differently, which affects both data filter configuration and target queries. Note the following:

• Oracle Database stores database, schema, and object names in uppercase unless explicitly quoted in the CREATE statement. For example, if you create mytable (no quotes), the Oracle data dictionary stores the table name as MYTABLE. If you quote the object name, the data dictionary preserves the case.

• Zero-ETL data filters are case sensitive and must match the exact case of object names as they appear in the Oracle data dictionary.

• Amazon Redshift queries default to lowercase object names unless explicitly quoted. For example, a query of MYTABLE (no quotes) searches for mytable.

Be mindful of the case differences when you create the Amazon Redshift filter and query the data.

Creating an uppercase integration

When you create a table without specifying the name in double quotes, the Oracle database stores the name in uppercase in the data dictionary. For example, you can create MYTABLE using any of the following SQL statements.

CREATE TABLE REINVENT.MYTABLE (id NUMBER PRIMARY KEY, description VARCHAR2(100));
CREATE TABLE reinvent.mytable (id NUMBER PRIMARY KEY, description VARCHAR2(100));
CREATE TABLE REinvent.MyTable (id NUMBER PRIMARY KEY, description VARCHAR2(100));
CREATE TABLE reINVENT.MYtabLE (id NUMBER PRIMARY KEY, description VARCHAR2(100));

Because you didn't quote the table name in the preceding statements, the Oracle database stores the object name in uppercase as MYTABLE.

To replicate this table to Amazon Redshift, you must specify the uppercase name in your data filter of your create-integration command. The Zero-ETL filter name and Oracle data dictionary name must match.

aws rds create-integration \
  --integration-name upperIntegration \
  --data-filter "include: ORCL.REINVENT.MYTABLE" \
...

By default, Amazon Redshift stores data in lowercase. To query MYTABLE in the replicated database in Amazon Redshift, you must quote the uppercase name MYTABLE so that it matches the case in the Oracle data dictionary.

SELECT * FROM targetdb1."REINVENT"."MYTABLE";

The following queries don't use the quoting mechanism. They all return an error because they search for an Amazon Redshift table named mytable, which uses the default lowercase name, but the table is named MYTABLE in the Oracle data dictionary.

SELECT * FROM targetdb1."REINVENT".MYTABLE;
SELECT * FROM targetdb1."REINVENT".MyTable;
SELECT * FROM targetdb1."REINVENT".mytable;

The following queries uses the quoting mechanism to specify a mixed case name. The queries all return an error because they search for an Amazon Redshift table that isn't named MYTABLE.

SELECT * FROM targetdb1."REINVENT"."MYtablE";
SELECT * FROM targetdb1."REINVENT"."MyTable";
SELECT * FROM targetdb1."REINVENT"."mytable";

Creating a lowercase integration

In the following alternative example, you use double quotes to store the table name in lowercase in the Oracle data dictionary. You create mytable as follows.

CREATE TABLE REINVENT."mytable" (id NUMBER PRIMARY KEY, description VARCHAR2(100));

The Oracle database stores the table name as mytable in lowercase. To replicate this table to Amazon Redshift, you must specify the lowercase name mytable in your Zero-ETL data filter.

aws rds create-integration \
  --integration-name lowerIntegration \
  --data-filter "include: ORCL.REINVENT.mytable" \
...

When you query this table in the replicated database in Amazon Redshift, you can specify the lowercase name mytable. The query succeeds because it searches for a table named mytable, which is the table name in the Oracle data dictionary.

SELECT * FROM targetdb1."REINVENT".mytable;

Because Amazon Redshift defaults to lowercase object names, the following queries also succeed in finding mytable.

SELECT * FROM targetdb1."REINVENT".MYtablE;
SELECT * FROM targetdb1."REINVENT".MYTABLE;
SELECT * FROM targetdb1."REINVENT".MyTable;

The following queries use the quoting mechanism for the object name. They all return an error because they search for an Amazon Redshift table whose name is different from mytable.

SELECT * FROM targetdb1."REINVENT"."MYTABLE";
SELECT * FROM targetdb1."REINVENT"."MyTable";
SELECT * FROM targetdb1."REINVENT"."MYtablE";

Create a table with a mixed-case integration

In the following example, you use double quotes to store the table name in lowercase in the Oracle data dictionary. You create MyTable as follows.

CREATE TABLE REINVENT."MyTable" (id NUMBER PRIMARY KEY, description VARCHAR2(100));

The Oracle database stores this table name as MyTable with mixed case. To replicate this table to Amazon Redshift, you must specify the mixed case name in the data filter.

aws rds create-integration \
  --integration-name mixedIntegration \
  --data-filter "include: ORCL.REINVENT.MyTable" \
...

When you query this table in the replicated database in Amazon Redshift, you must specify the mixed case name MyTable by quoting the object name.

SELECT * FROM targetdb1."REINVENT"."MyTable";

Because Amazon Redshift defaults to lowercase object names, the following queries don't find the object because they are searching for the lowercase name mytable.

SELECT * FROM targetdb1."REINVENT".MYtablE;
SELECT * FROM targetdb1."REINVENT".MYTABLE;
SELECT * FROM targetdb1."REINVENT".mytable;

Note

You can't use regular expressions in the filter value for database name, schema, or table name in RDS for Oracle integrations.

Adding data filters to an integration

You can configure data filtering using the AWS Management Console, the AWS CLI, or the Amazon RDS API.

Important

If you add a filter after you create an integration, Amazon RDS treats it as if it always existed. It removes any data in the target data warehouse that doesn’t match the new filtering criteria and resynchronizes all affected tables.

RDS console

To add data filters to a zero-ETL integration

1. Sign in to the AWS Management Console and open the Amazon RDS console at https://console.aws.amazon.com/rds/.

2. In the navigation pane, choose Zero-ETL integrations. Select the integration that you want to add data filters to, and then choose Modify.

3. Under Source, add one or more Include and Exclude statements.

   The following image shows an example of data filters for a MySQL integration:

   

4. When you're satisfied with the changes, choose Continue and Save changes.

AWS CLI

To add data filters to a zero-ETL integration using the AWS CLI, call the modify-integration command. In addition to the integration identifier, specify the --data-filter parameter with a comma-separated list of Include and Exclude Maxwell filters.

The following example adds filter patterns to my-integration.

For Linux, macOS, or Unix:

aws rds modify-integration \
    --integration-identifier my-integration \
    --data-filter 'include: foodb.*, exclude: foodb.tbl, exclude: foodb./table_\d+/'

For Windows:

aws rds modify-integration ^
    --integration-identifier my-integration ^
    --data-filter 'include: foodb.*, exclude: foodb.tbl, exclude: foodb./table_\d+/'

RDS API

To modify a zero-ETL integration using the RDS API, call the ModifyIntegration operation. Specify the integration identifier and provide a comma-separated list of filter patterns.

Removing data filters from an integration

When you remove a data filter from an integration, Amazon RDS reevaluates the remaining filters as if the removed filter never existed. It then replicates any previously excluded data that now meets the criteria into the target data warehouse. This triggers a resynchronization of all affected tables.