This page explains how fine-grained access control works with Spanner models.
For fine-grained access control users, you can control access to MODEL entities with the following privilege:
Grant
EXECUTE
on the model to allow machine learning functions to use it.GoogleSQL
GRANT EXECUTE ON MODEL MODEL_NAME TO ROLE ROLE_NAME;
INFORMATION_SCHEMA
views for models
The following views show the database roles and privileges information for models:
- GoogleSQL-dialect databases:
INFORMATION_SCHEMA.MODEL_PRIVILEGES
The rows in this view are filtered based on the current database role's privileges on models. This ensures that principals can view only the roles, privileges, and models that they have access to.
Row filtering also applies to the following model-related views:
GoogleSQL
The system role spanner_info_reader
and its members always see an unfiltered
INFORMATION_SCHEMA
.