Skip to Content
FeaturesIntegrationsDatabases & Warehouses

Databases & Warehouses

First‑class support for managing deletions in data platforms such as Snowflake, BigQuery, and Postgres. Forget API executes provider‑appropriate routines that align with your business logic and compliance needs.

Overview

Warehouse/Database integrations execute data changes directly against your data platform using a secure Account, discovered schema, and configurable rules. You define what to change and when, and Forget API generates and runs the appropriate SQL.

Prerequisites

  • An authenticated Account for your warehouse/database (e.g., Snowflake role with private key, BigQuery service account, Postgres user)
  • Proper privileges for the service identity to read schema and perform configured actions

Supported patterns

  • Hard delete: permanently remove rows by identifier
  • Soft delete: set a deletion flag or timestamp for down‑stream masking
  • Pseudonymization/masking: redact or hash PII fields
  • Scheduled jobs: defer heavy operations to maintenance windows

Connection & schema discovery

  • Connect an Account with the necessary credentials and role
  • Forget API runs an information schema fetch to list available tables and columns (requires metadata read permissions)
  • The schema view powers the rule builder in the UI

Configuration (rule builder)

  1. Choose tables
  • Select one or more tables you want to process for the deletion request
  1. Identify matching rows
  • Pick one or more identifier columns that are compared to the Request identifier
  • Set a matching strategy: any or all of the identifier columns must match
  1. Define actions per rule
  • deleteValue: set/clear a column value (requires column to be nullable)
  • hashValue: deterministically hash a column value
  • maskValue: replace a column value with a masked/redacted form
  • deleteRow: remove matching rows entirely

Notes on target columns

  • For deleteValue, only nullable columns are selectable
  • For hashValue and maskValue, only updatable columns are selectable
  1. Chain multiple actions
  • Add multiple actions per rule; actions run in order

Execution model

Forget API batches warehouse/database deletions and executes them automatically every 12 hours. Within each batch window, requests are grouped by workspace, tables, and actions to generate the fewest queries possible while honoring your rules.

  • Batch cadence: every 12 hours
  • Grouping: by workspace and compatible rule sets (tables, identifier columns, actions)
  • Optimization: multiple identifiers are handled per generated statement when safe

Results, row counts, and any provider messages are recorded in audit logs.

Permissions

Grant least‑privilege access to the service identity:

  • Metadata read to discover schema (e.g., access to INFORMATION_SCHEMA)
  • USAGE/ROLE privileges as required by your platform (database, schema, warehouse/project)
  • SELECT/UPDATE/DELETE on the specific tables and columns affected by your rules

Best practices

  • Use least‑privilege service roles limited to required tables
  • Keep a single source of truth for identifier mapping
Last updated on
WebhooksProductivity