Doris

Query existing Paimon tables through a Doris external catalog. Use Connecting Engines to identify the catalog backend and warehouse first.

Version

Select a Doris release that supports your catalog backend and Paimon table features. The Doris Paimon catalog documentation provides version-specific settings. The REST catalog example below requires Doris 3.1 or later.

This integration reads Paimon tables. An external catalog exposes existing tables without copying them into Doris; it does not enable writes to Paimon.

Prerequisites

Prepare an existing Paimon table and configure metastore and storage access for the Doris processes that perform metadata and data reads. The examples below use placeholder hosts and credentials; replace them with your deployment's settings.

Create Paimon Catalog

Choose one catalog definition for the backend used by the writer. HDFS examples assume the cluster's Hadoop access is already configured.

Filesystem Catalog on HDFS

CREATE CATALOG paimon_hdfs PROPERTIES (
    'type' = 'paimon',
    'paimon.catalog.type' = 'filesystem',
    'warehouse' = 'hdfs://namenode:8020/warehouse/paimon',
    'hadoop.username' = 'hadoop'
);

Filesystem Catalog on OSS

CREATE CATALOG paimon_oss PROPERTIES (
    'type' = 'paimon',
    'paimon.catalog.type' = 'filesystem',
    'warehouse' = 'oss://paimon-bucket/warehouse',
    'oss.endpoint' = 'oss-cn-beijing.aliyuncs.com',
    'oss.access_key' = '<access-key-id>',
    'oss.secret_key' = '<access-key-secret>'
);

Hive Metastore Catalog

CREATE CATALOG paimon_hms PROPERTIES (
    'type' = 'paimon',
    'paimon.catalog.type' = 'hms',
    'warehouse' = 'hdfs://namenode:8020/warehouse/paimon',
    'hive.metastore.uris' = 'thrift://metastore:9083',
    'hadoop.username' = 'hadoop'
);

DLF 1.0 Catalog

CREATE CATALOG paimon_dlf PROPERTIES (
    'type' = 'paimon',
    'paimon.catalog.type' = 'dlf',
    'warehouse' = 'oss://paimon-bucket/warehouse',
    'dlf.proxy.mode' = 'DLF_ONLY',
    'dlf.uid' = '<account-id>',
    'dlf.region' = 'cn-beijing',
    'dlf.access_key' = '<access-key-id>',
    'dlf.secret_key' = '<access-key-secret>'
);

DLF REST Catalog

For Doris 3.1+, configure the REST backend explicitly. Here, warehouse is the DLF catalog name, rather than an object-storage path.

CREATE CATALOG dlf_paimon_rest PROPERTIES (
    'type' = 'paimon',
    'paimon.catalog.type' = 'rest',
    'uri' = 'http://cn-beijing-vpc.dlf.aliyuncs.com',
    'warehouse' = '<catalog-name>',
    'paimon.rest.token.provider' = 'dlf',
    'paimon.rest.dlf.access-key-id' = '<access-key-id>',
    'paimon.rest.dlf.access-key-secret' = '<access-key-secret>'
);

See the Doris catalog examples for storage authentication and differences between Doris releases.

Access Paimon Catalog

Query a fully qualified table name:

SELECT * FROM paimon_hdfs.paimon_db.paimon_table LIMIT 10;

Alternatively, select the catalog and database for the session:

SWITCH paimon_hdfs;
USE paimon_db;
SELECT * FROM paimon_table LIMIT 10;

Query Optimization

Primary-key tables can require merging multiple row versions. Review the reader capabilities in your Doris release before selecting a table mode.

• Read optimized: reading compacted base data can reduce merge work, but a read-optimized view can omit recent changes.
• Deletion vectors: readers must apply deletion vectors to suppress obsolete rows. See Merge On Write for the Paimon configuration and read semantics, and check Doris support before enabling it.

Type Mapping

This is a summary of common Paimon-to-Doris read mappings. Check the Doris type mapping for the installed version; binary and timestamp mappings can depend on release and catalog options.

Paimon type	Doris type
BOOLEAN	BOOLEAN
TINYINT, SMALLINT, INT, BIGINT	Corresponding integer type
FLOAT, DOUBLE	FLOAT, DOUBLE
DECIMAL(p, s)	DECIMAL(p, s)
CHAR, VARCHAR	STRING
BINARY, VARBINARY	See the version-specific binary mapping
DATE	DATE
TIMESTAMP, TIMESTAMP WITH LOCAL TIME ZONE	See the version-specific timestamp mapping and precision limits
ARRAY	ARRAY
MAP	MAP
ROW	STRUCT

Next Steps

Use the Doris guide for time-travel and system-table syntax supported by your release. For connection or visibility problems, follow Connecting Engines.