Sink data from RisingWave to PostgreSQL

This guide will show you how to sink data from RisingWave to PostgreSQL using the JDBC connector. The sink parameters are similar to those for other JDBC-available databases, such as MySQL. However, we will cover the configurations specific to PostgreSQL and how to verify that data is successfully sunk.

You can test out this process on your own device by using the postgres-sink demo in the integration_test directory of the RisingWave repository.

Set up a PostgreSQL database

AWS RDS

Set up a PostgreSQL RDS instance on AWS

Here we will use a standard class instance without Multi-AZ deployment as an example.

1. Log in to the AWS console. Search “RDS” in services and select the RDS panel.
2. Create a database with PostgreSQL as the Engine type. We recommend setting up a username and password or using other security options.
3. When the new instance becomes available, click on its panel.
4. From the Connectivity panel, we can find the endpoint and connection port information.

Connect to the RDS instance from Postgres

Now we can connect to the RDS instance. Make sure you have installed psql on your local machine, and start a psql prompt. Fill in the endpoint, the port, and login credentials in the connection parameters.

psql --host = pg-to-rw.xxxxxx.us-east-1.rds.amazonaws.com --port=5432 --username=awsuser --password

For more login options, refer to the RDS connection guide.

Self-hosted

Launch and set up PostgreSQL

To install PostgreSQL locally, see their download options.

If you are using the demo version, connect to PostgreSQL with the following command. Ensure that all other programs are disconnected from port 5432.

psql postgresql://myuser:123456@127.0.0.1:5432/mydb

Ensure that the Postgres user is granted the following privileges on the used table with the following SQL query.

GRANT SELECT, INSERT, UPDATE, DELETE ON [table_name] TO [username];

Create a table in PostgreSQL

Use the following query to set up a table in PostgreSQL. We will sink to this table from RisingWave.

CREATE TABLE target_count (
  target_id VARCHAR(128) PRIMARY KEY,
  target_count BIGINT
);

Set up RisingWave

Install and launch RisingWave

To install and start RisingWave locally, see the Get started guide. We recommend running RisingWave locally for testing purposes.

Notes about running RisingWave from binaries

If you are running RisingWave locally from binaries and intend to use the native CDC source connectors or the JDBC sink connector, make sure you have JDK 11 or later versions installed in your environment.

Create a sink

Syntax

CREATE SINK [ IF NOT EXISTS ] sink_name
[FROM sink_from | AS select_query]
WITH (
   connector = 'jdbc' | 'postgres',
   field_name = 'field', ...
);

Parameters (JDBC)

Parameter or clause	Description
sink_name	Name of the sink to be created.
sink_from	A clause that specifies the direct source from which data will be output. sink_from can be a materialized view or a table. Either this clause or a SELECT query must be specified.
AS select_query	A SELECT query that specifies the data to be output to the sink. Either this query or a FROM clause must be specified. See SELECT for the syntax and examples of the SELECT command.
connector	Sink connector should be jdbc. To switch from jdbc to postgres, set stream_switch_jdbc_pg_to_native = true under [streaming.developer].
jdbc.url	Required. The JDBC URL of the destination database necessary for the driver to recognize and connect to the database.
user	The user name for the database connection.
password	The password for the database connection.
jdbc.query.timeout	Specifies the timeout for the operations to downstream. If not set, the default is 60s.
jdbc.auto.commit	Controls whether to automatically commit transactions for JDBC sink. If not set, the default is false.
table.name	Required. The table in the destination database you want to sink to.
schema.name	The schema in the destination database you want to sink to. The default value is public.
type	Sink data type. Supported types: append-only: Sink data as INSERT operations. upsert: Sink data as UPDATE, INSERT and DELETE operations.
primary_key	Required if type is upsert. The primary key of the sink, which should match the primary key of the downstream table.
tcp.keepalive.enable	Enables TCP keepalive for the connection to the downstream database. If not set, the default is false.
tcp.keepalive.idle	The time (in seconds) the connection must be idle before the first keepalive probe is sent. Takes effect only when tcp.keepalive.enable is true. If not set, the default is 600 (10 minutes).
tcp.keepalive.interval	The time (in seconds) between consecutive keepalive probes. Takes effect only when tcp.keepalive.enable is true. If not set, the default is 10.
tcp.keepalive.count	The number of unacknowledged keepalive probes before the connection is considered dead. Takes effect only when tcp.keepalive.enable is true. If not set, the default is 3.

Parameters (Postgres Native)

RisingWave introduced the native Postgres sink connector in v2.2.0, and the JDBC sink connector for Postgres will be deprecated in a future release. Existing JDBC sinks will be maintained in a backwards-compatible way.

There are some known issues for native Postgres Sink connector. If you encounter these, we recommend using the JDBC sink first. You can try switching your JDBC sinks to the native Postgres sink in place by setting stream_switch_jdbc_pg_to_native = true under [streaming.developer].

Parameter or clause	Description
sink_name	Name of the sink to be created.
sink_from	A clause that specifies the direct source from which data will be output. sink_from can be a materialized view or a table. Either this clause or a SELECT query must be specified.
AS select_query	A SELECT query that specifies the data to be output to the sink. Either this query or a FROM clause must be specified. See SELECT for the syntax and examples of the SELECT command.
connector	Sink connector must be postgres.
user	The user name for the database connection.
password	The password for the database connection.
database	Required. The database in the destination database you want to sink to.
table	Required. The table in the destination database you want to sink to.
schema	The schema in the destination database you want to sink to. The default value is public.
type	Sink data type. Supported types: append-only: Sink data as INSERT operations. upsert: Sink data as UPDATE, INSERT and DELETE operations.
primary_key	Required if type is upsert. The primary key of the sink, which should match the primary key of the downstream table.
ssl_mode	The ssl.mode parameter determines the level of SSL/TLS encryption for secure communication with Postgres. Accepted values are disabled, preferred, required, verify-ca, and verify-full. The default value is disabled. When set to required, it enforces TLS for establishing a connection; When set to verify-ca, it verifies that the server is trustworthy by checking the certificate chain up to the root certificate stored on the client; When set to verify-full, it verifies the certificate and also ensures the server hostname matches the name in the certificate.
ssl_root_cert	Specify the root certificate secret. You must create secret first and then use it here.
tcp.keepalive.enable	Enables TCP keepalive for the connection to the downstream database. If not set, the default is false.
tcp.keepalive.idle	The time (in seconds) the connection must be idle before the first keepalive probe is sent. Takes effect only when tcp.keepalive.enable is true. If not set, the default is 600 (10 minutes).
tcp.keepalive.interval	The time (in seconds) between consecutive keepalive probes. Takes effect only when tcp.keepalive.enable is true. If not set, the default is 10.
tcp.keepalive.count	The number of unacknowledged keepalive probes before the connection is considered dead. Takes effect only when tcp.keepalive.enable is true. If not set, the default is 3.

Sink data from RisingWave to PostgreSQL

Create source and materialized view

You can sink data from a table or a materialized view in RisingWave to PostgreSQL.

For demonstration purposes, we’ll create a source and a materialized view, and then sink data from the materialized view. If you already have a table or materialized view to sink data from, you don’t need to perform this step.

Run the following query to create a source to read data from a Kafka broker.

CREATE SOURCE user_behaviors (
    user_id VARCHAR,
    target_id VARCHAR,
    target_type VARCHAR,
    event_timestamp TIMESTAMPTZ,
    behavior_type VARCHAR,
    parent_target_type VARCHAR,
    parent_target_id VARCHAR
) WITH (
    connector = 'kafka',
    topic = 'user_behaviors',
    properties.bootstrap.server = 'message_queue:29092',
    scan.startup.mode = 'earliest'
) FORMAT PLAIN ENCODE JSON;

Next, we will create a materialized view that queries the number of targets for each target_id. Note that the materialized view and the target table share the same schema.

CREATE MATERIALIZED VIEW target_count AS
SELECT
    target_id,
    COUNT(*) AS target_count
FROM
    user_behaviors
GROUP BY
    target_id;

Sink from RisingWave

Use the following query to sink data from the materialized view to the target table in PostgreSQL. Ensure that the jdbc_url is accurate and reflects the PostgreSQL database that you are connecting to. See CREATE SINK for more details.

CREATE SINK target_count_postgres_sink FROM target_count WITH (
    connector = 'jdbc',
    jdbc.url = 'jdbc:postgresql://postgres:5432/mydb',
    user = 'myuser',
    password = '123456',
    table.name = 'target_count',
    type = 'upsert',
    primary_key = 'target_id'
);

Verify update

To ensure that the target table has been updated, query from target_count in PostgreSQL.

SELECT * FROM target_count
LIMIT 10;

Data type mapping

For the PostgreSQL data type mapping table, see the Data type mapping table under the Ingest data from PostgreSQL CDC topic.

Additional notes regarding sinking data to PostgreSQL:

• A varchar column in RisingWave can be sinked to a uuid column in Postgres.
• Only one-dimensional arrays in RisingWave can be sinked to PostgreSQL.
• For array type, we only support smallint, integer, bigint, real, double precision, and varchar type now.