ClickHouse External Table

Since Proton v1.4.2, it added the support to read or write ClickHouse tables. This unlocks a set of new use cases, such as

• Use Proton to efficiently process real-time data in Kafka/Redpanda, apply flat transformation or stateful aggregation, then write the data to the local or remote ClickHouse for further analysis or visualization.
• Enrich the live data with the static or slow-changing data in ClickHouse. Apply streaming JOIN.
• Use Proton to query historical or recent data in ClickHouse

This integration is done by introducing a new concept in Proton: "External Table". Similar to External Stream, there is no data persisted in Proton. However, since the data in ClickHouse is in the form of table, not data stream, so we call this as External Table. In the roadmap, we will support more integrations by introducing other types of External Table.

Demo Video

This video demonstrates how to read live data from Redpanda, apply stream processing and send results to ClickHouse.

CREATE EXTERNAL TABLE

Syntax

CREATE EXTERNAL TABLE name
SETTINGS type='clickhouse',
         address='..',
         user='..',
         password='..',
         database='..',
         secure=true|false,
         table='..';

The required settings are type and address. For other settings, the default values are

• 'default' for user
• '' (empty string) for password
• 'default' for database
• 'false' for secure
• If you omit the table name, it will use the name of the external table

You don't need to specify the columns, since the table schema will be fetched from the ClickHouse server.

Once the external table is created successfully, you can run the following SQL to list the columns:

DESCRIBE name

info

The data types in the output will be Proton data types, such as uint8, instead of ClickHouse type UInt8. Proton maintains a mapping for those types. Learn more.

You can define the external table and use it to read data from the ClickHouse table, or write to it.

Connect to a local ClickHouse

Example SQL to connect to a local ClickHouse server without password:

CREATE EXTERNAL TABLE ch_local
SETTINGS type='clickhouse',
         address='localhost:9000',
         table='events'

Connect to ClickHouse Cloud

Example SQL to connect to ClickHouse Cloud:

CREATE EXTERNAL TABLE ch_cloud
SETTINGS type='clickhouse',
         address='abc.clickhouse.cloud:9440',
         user='default',
         password='..',
         secure=true,
         table='events';

Connect to Aiven for ClickHouse

Example SQL to connect to Aiven for ClickHouse:

CREATE EXTERNAL TABLE ch_aiven
SETTINGS type='clickhouse',
         address='abc.aivencloud.com:28851',
         user='avnadmin',
         password='..',
         secure=true,
         table='events';

Read data from ClickHouse

Once the external table is created successfully, it means Proton can connect to the ClickHouse server and fetch the table schema.

You can query it via the regular select .. from table_name.

warning

Please note, in the current implementation, all rows will be fetched from ClickHouse to Proton, with the selected columns. Then Proton applies the SQL functions and LIMIT n locally. It's not recommended to run SELECT * for a large ClickHouse table.

Also note, use the Proton function names when you query the external table, such as to_int, instead of ClickHouse's naming convension, e.g. toInt. In current implementation, the SQL functions are applied in Proton engine. We plan to support some function push-down to ClickHouse in future versions.

Limitations:

1. tumble/hop/session/table functions are not supported for External Table (coming soon)
2. scalar or aggregation functions are performed by Proton, not the remote ClickHouse
3. LIMIT n is performed by Proton, not the remote ClickHouse

Write data to ClickHouse

You can run regular INSERT INTO to add data to ClickHouse table. However it's more common to use a Materialized View to send the streaming SQL results to ClickHouse.

Say you have created an external table ch_table. You can create an materialized view to read Kafka data(via an external stream) and transform/aggregate the data and send to the external table:

-- setup the ETL pipeline via a materialized view
CREATE MATERIALIZED VIEW mv INTO ch_table AS
    SELECT now64() AS _tp_time,
           raw:requestedUrl AS url,
           raw:method AS method,
           lower(hex(md5(raw:ipAddress))) AS ip
    FROM kafka_events;

Supported data types

All ClickHouse data types are supported in the external table. While reading or writing data, Proton applies a data type mapping, such as converting Proton's uint8 to ClickHouse's UInt8. If you find anything wrong with the data type, please let us know.