Trino

Use the separately distributed Paimon connector to query tables from Trino. The examples on this page describe the Trino 440 connector. For other versions, follow the matching revision of Apache Paimon Trino.

Version

Match the Trino server version, connector artifact, and Java runtime. The connector repository has its own release cycle and Paimon dependency; its default branch can target a different Trino and Java version from the examples here.

The write examples below require a connector with write support and a supported table layout. They do not imply that every historical Trino connector supports these operations.

Installation

Obtain the Plugin

Use the download page to locate the Trino 440 plugin archive. For other versions, use the connector repository. The plugin is a distribution containing its dependencies, rather than a single jar to place on the general Trino classpath.

To build from source, check out a connector revision matching the Trino server and use the Java version required by that revision's pom.xml. Run its documented build; for revisions using the Maven build, the command is:

mvn clean install -DskipTests

Locate the plugin archive in that revision's build output. Artifact names and module paths vary across revisions. Use the connector artifact's actual version rather than substituting the Paimon version shown by this documentation.

Install Paimon Connector

For a Trino 440 distribution named paimon-trino-440-<connector-version>-plugin.tar.gz, extract it under the Trino plugin directory on the coordinator and every worker:

tar -zxf paimon-trino-440-<connector-version>-plugin.tar.gz -C "${TRINO_HOME}/plugin"

For the Trino 440 connector running on JDK 21, include these entries in etc/jvm.config:

--add-opens=java.base/sun.nio.ch=ALL-UNNAMED
--add-opens=java.base/java.nio=ALL-UNNAMED

Restart Trino after installing the plugin and completing the catalog configuration.

Configure Paimon Catalog

Create etc/catalog/paimon.properties on the Trino nodes. The file name registers the SQL catalog as paimon:

connector.name=paimon
warehouse=hdfs://namenode:8020/warehouse/paimon

This example uses a filesystem catalog and assumes HDFS access is configured. For a local single-node experiment, warehouse=file:/tmp/warehouse is sufficient. Use shared storage for a distributed deployment.

Filesystem

The Trino 440 connector uses Trino's filesystem integration. Configure storage access before querying a table. For HDFS, provide HADOOP_HOME, HADOOP_CONF_DIR, or the connector's hadoop-conf-dir property as appropriate for the deployment.

For Hadoop-backed object storage, supply the filesystem configuration through:

hive.config.resources=/path/to/core-site.xml

Make the configuration file and required filesystem libraries available on all nodes that access the warehouse. Follow the filesystem instructions for your Trino and connector versions; properties from newer Trino releases may differ.

Connector revisions using io.trino.hadoop:hadoop-apache can override that dependency when a specific Hadoop distribution is required. For example, if supported by the selected revision:

mvn clean install -DskipTests -Dhadoop.apache.version=3.3.5-1

Kerberos

For the Trino 440 connector's Kerberos login, set these catalog properties:

security.kerberos.login.principal=hadoop-user@EXAMPLE.COM
security.kerberos.login.keytab=/etc/trino/hdfs.keytab

Distribute the keytab to every Trino node that needs it and configure access for the service user.

Temporary Directory

Paimon extracts jars for code generation into the JVM temporary directory. Choose a writable location that is not removed by periodic cleanup while Trino is running. Set this JVM option in etc/jvm.config on each node:

-Djava.io.tmpdir=/path/to/trino-tmp

Create Schema

The following examples form one sequence in the paimon catalog:

CREATE SCHEMA paimon.test_db;

Create Table

Create a primary-key table with a fixed bucket count. Include the partition column in the primary key:

CREATE TABLE paimon.test_db.orders (
    order_key bigint,
    order_status varchar,
    total_price decimal(18,4),
    order_date date
)
WITH (
    file_format = 'ORC',
    primary_key = ARRAY['order_key', 'order_date'],
    partitioned_by = ARRAY['order_date'],
    bucket = '2',
    bucket_key = 'order_key',
    changelog_producer = 'input'
);

Insert

The Trino 440 connector supports inserts into primary-key tables with fixed buckets and non-primary-key tables with bucket = -1.

INSERT INTO paimon.test_db.orders
    (order_key, order_status, total_price, order_date)
VALUES (1, 'NEW', DECIMAL '19.9900', DATE '2024-01-01');

Query

SELECT * FROM paimon.test_db.orders;

Add Column

Continue with the existing table; there is no need to create it again:

ALTER TABLE paimon.test_db.orders ADD COLUMN shipping_address varchar;

Query with Time Traveling

Use a retained snapshot or tag. The timestamp and identifiers below are examples; replace them with values from the table's history.

-- Select the snapshot at the specified timestamp.
SELECT * FROM paimon.test_db.orders
FOR TIMESTAMP AS OF TIMESTAMP '2024-01-01 00:00:00 Asia/Shanghai';

-- Select snapshot 1 if it is still retained.
SELECT * FROM paimon.test_db.orders FOR VERSION AS OF 1;

-- Select an existing tag.
SELECT * FROM paimon.test_db.orders FOR VERSION AS OF 'my-tag';

Numeric tag names

A numeric tag name takes precedence over a matching snapshot ID when supplied as a string. If tag '1' points to snapshot 2, FOR VERSION AS OF '1' reads snapshot 2. Prefer descriptive tag names to avoid ambiguity.

See Snapshot Management and Tags for retention and tag creation.

Type Mapping

Common read mappings for the connector are summarized below using SQL type names. Check the matching connector revision for precision limits and unsupported types.

Paimon type	Trino type
BOOLEAN	BOOLEAN
TINYINT, SMALLINT, INT, BIGINT	TINYINT, SMALLINT, INTEGER, BIGINT
FLOAT, DOUBLE	REAL, DOUBLE
DECIMAL(p, s)	DECIMAL(p, s)
CHAR(n)	CHAR(n)
VARCHAR(n), STRING	VARCHAR(n), VARCHAR
VARBINARY	VARBINARY
DATE	DATE
TIMESTAMP	TIMESTAMP
TIMESTAMP WITH LOCAL TIME ZONE	TIMESTAMP WITH TIME ZONE
ARRAY, MAP, ROW	ARRAY, MAP, ROW

Next Steps

Use Connecting Engines for catalog and storage checks. Report connector-specific problems to Apache Paimon Trino, including the Trino, Java, and connector versions and the table options.