Oracle Analytics and AI

Sarma NS-Oracle

1. Check if the database supports the VECTOR data type

SELECT banner_full FROM v$version;
SELECT type_name FROM all_types WHERE type_name = 'VECTOR';

Note: Oracle Database 23ai / 23c supports the VECTOR data type. Older versions usually do not.

1. Check whether any table has VECTOR columns

Run this as a user with dictionary access:

SELECT owner,table_name,column_name,data_type FROM all_tab_columns WHERE data_type = 'VECTOR' ORDER BY owner, table_name, column_id;

If it returns rows, the database has tables with vector columns.

For current schema only:

SELECT table_name,column_name,data_type FROM user_tab_columns WHERE data_type = 'VECTOR' ORDER BY table_name, column_id;

1. Check for vector indexes

Oracle vector search often uses vector indexes. Check:

SELECT owner,index_name,table_name,index_type FROM all_indexes WHERE index_type LIKE '%VECTOR%' ORDER BY owner, table_name, index_name;

Or for your schema:

SELECT index_name,table_name,index_type FROM user_indexes WHERE index_type LIKE '%VECTOR%';

1. Check a specific table

DESC your_table_name;
Look for columns with type like:

VECTOR
VECTOR(1536, FLOAT32)

Example query:

SELECT column_name,data_type,data_type_mod,data_type_owner FROM user_tab_columns WHERE table_name = UPPER('your_table_name');

1. Check whether vector packages/features are available
   SELECT object_name, object_type FROM all_objects WHERE object_name LIKE '%VECTOR%' ORDER BY object_type, object_name;

For Oracle AI Vector Search, you may see objects/packages related to vector operations depending on version and privileges.

Most direct check:

SELECT owner, table_name, column_name FROM all_tab_columns WHERE data_type = 'VECTOR';