Skip to content

Migration · By Rivtor · Updated

Supabase to PostgreSQL migration: database, auth and storage boundaries

A Supabase migration must separate PostgreSQL data from Supabase-specific authentication, storage, realtime and API integrations. A standard PostgreSQL destination may hold relational data but does not automatically replace those services or their client contracts.

Compare Rivtor plans for your workload →

Inventory more than tables

Search the application for Supabase client calls, authentication helpers, storage buckets, realtime subscriptions and generated API usage. Also inventory database policies, extensions, functions, triggers and role assumptions. A table export does not include the behaviour of every surrounding service.

If the application relies on those features, the appropriate decision can be to keep Supabase while moving only another service. Do not present a new PostgreSQL instance as a drop-in replacement for the entire platform.

Classify the compatibility work

DependencyQuestion before migration
Relational tablesAre schema, types, extensions and data portable?
AuthenticationWhat verifies identities and existing sessions?
StorageHow are objects, policies and URLs retained?
RealtimeWhat supplies the existing subscription contract?
Database roles/policiesWho executes queries and enforces access?

Rivtor’s data services document PostgreSQL and object storage boundaries. They do not establish feature parity with Supabase Auth, Realtime or every extension. Required application feature changes fall outside a host-only migration and need their own design.

Rehearse only the supported subset

For an application using ordinary SQL through its own backend, test schema compatibility in a disposable destination. Check extension availability and restore privileges first. Restore a reviewed export and verify representative queries, constraints and sequences.

For direct browser access through Supabase APIs, do not point the browser at a raw database URL. Retain the existing service or design an authorized application API separately. Never expose privileged service keys to make a migration appear complete.

Decide whether the move is justified

Compare engineering effort and ongoing service requirements with the expected operational benefit. Define data freeze, final export and rollback only after compatibility is established. Verify two-account access controls and file ownership, not just row counts.

A partial migration can be a valid architecture if the network path and operational boundaries are understood. An honest migration guide should state the features that remain elsewhere instead of implying a complete replacement from a successful SQL import.

References