Top
Enterprise Postgres 17 SP1 Knowledge DataManagement FeatureUser's Guide

3.5.1 Encrypting Vector Data for Semantic Text Search

Several database objects are created for semantic text search, and if you want to encrypt these database objects together, place the entire database in an encryption tablespace.

Example) Encrypting the entire database

postgres=# CREATE DATABASE rag_database TABLESPACE = encrypted_tablespace;

rag_database=> CREATE EXTENSION pgx_vectorizer;

To encrypt database objects for semantic text search when a single tablespace cannot be used per database, temporarily change the default tablespace to an encrypted tablespace before defining vectorization.

Indexes for vector data are created when the index creation conditions are met. To encrypt an index, change the tablespace after index creation, or create the index manually without specifying an index in the create_vectorizer function. For information about tables containing vector data created by this feature, refer to "3.11.6 Tables/Views Created by Semantic Text Search and Automatic Vectorization Feature".