Skip to main content
This connector requires the IBM Data Server Driver for ODBC and CLI on your deployment. Validate connectivity against your server before production use. IBM i is not supported by this connector.

​
Prerequisites

  • A reachable DRDA endpoint and a database name (LUW) or DDF location name (z/OS).
  • A dedicated account with SELECT access to the intended tables and read access to the system catalogs. Remove write privileges and access to routines that modify data. The connector requests IBM’s ReadOnlyConnection mode for reads; database grants remain required. The separate SQL_ATTR_ACCESS_MODE attribute is advisory.
  • The IBM driver, its runtime libraries, and any required Db2 Connect entitlement. Mainframe licensing depends on the server and client installation; confirm it with your IBM administrator.
The deployment operator installs the complete IBM ODBC/CLI driver tree and sets DB2_ODBC_DRIVER to the absolute path of libdb2.so in the compute container. Keep IBM’s dependencies and any license files in their supported locations. Nix builds use db2_odbc; Docker builds enable ODBC through teradata_odbc. Both require unixODBC. The connector does not redistribute IBM binaries or licenses.

​
Add the connector

  1. Open Connectors, add a database connector, and select IBM Db2.
  2. Select LUW or Mainframe (z/OS) to choose the matching system catalog.
  3. Enter the host, DRDA listener port, database or location, username, and password. Common non-TLS ports are 50000 for LUW and 446 for z/OS; use the listener port provided by your administrator.
  4. Enter the schema names, separated by commas. Match their catalog case, usually uppercase. This list limits schema discovery; database grants control query access.
  5. Keep Use TLS enabled for certificate and hostname validation. Configure private CA trust in the IBM driver when needed.
  6. Test the connection, then save it.

​
Queries and metadata

Tables, views, columns, types, and nullability are discovered from SYSCAT on LUW and SYSIBM on z/OS. Table previews use schema-qualified names and FETCH FIRST limits. Passwords use the existing connector encryption and redaction. Write native Db2 SQL. Read-only query verification accepts a single SELECT or set query, checks nested writes, and rejects syntax it cannot verify. Common trailing Db2 isolation/read-only clauses are supported. Table previews cast DECIMAL/NUMERIC columns to text using catalog types, preserving their exact digits. Native queries use the same upstream ODBC driver as Teradata, which returns uncast DECIMAL/NUMERIC values as floating point. For exact values in native queries, cast the final decimal expression, for example SELECT CAST(SUM(AMOUNT) AS VARCHAR(64)) AS TOTAL FROM CLAIMS.PAYMENTS. Keep arithmetic in Db2 before the final cast. Character values use the upstream driver mappings; binary values that are not valid UTF-8 are returned with a base64: prefix. Results exceeding 50,000 rows or 64 MiB return an error asking for a smaller result. Automatic PostgreSQL-to-Db2 conversion, automatic sample generation, primary-key discovery, foreign-key discovery, and automatic row counts are not supported in this version.

​
Troubleshooting

  • Driver unavailable: verify the absolute driver path and that the compute process can load all IBM runtime libraries.
  • SQL1598N or a license error: ask your IBM administrator to verify the Db2 Connect entitlement and installation.
  • Certificate error: verify the TLS port, certificate hostname, and IBM driver trust configuration.
  • No tables: verify the platform, schema spelling, and catalog and table privileges.
See IBM’s ODBC/CLI connection guide and connection attribute reference.