How to Migrate from Postgres to ParadeDB in PostgreSQL

Why migrate from PostgreSQL to ParadeDB?

ParadeDB keeps PostgreSQL compatibility while adding fast vector search, hybrid filtering, and scalable indexing. Migrating lets you reuse schemas, clients, and SQL while unlocking AI-ready search.

What are the prerequisites?

Ensure PostgreSQL ≥14, superuser rights, and ParadeDB binaries installed. Always back up with pg_dump before proceeding.

How to install the ParadeDB extension?

Connect as a superuser and run CREATE EXTENSION IF NOT EXISTS paradedb;. Verify with \dx.

How do I migrate a full schema?

Run SELECT paradedb.migrate_from_postgres('public');.The function copies every table, constraint, and index from public into a new ParadeDB-managed schema with the same name.

How do I migrate selected tables?

Pass named parameters such as include_tables, exclude_tables, move_indexes, and dry_run to tailor the migration.

Selective migration example

SELECT paradedb.migrate_from_postgres(source_schema=>'public', include_tables=>ARRAY['Customers','Orders'], move_indexes=>TRUE);

How to verify the migration?

Compare row counts: SELECT 'Customers' AS t, (SELECT COUNT(*) FROM Customers) AS pg, (SELECT COUNT(*) FROM paradedb.Customers) AS pddb;.Use checksums for large tables.

Can I roll back?

Drop the ParadeDB schema DROP SCHEMA paradedb CASCADE; or restore from your pre-migration backup.

Best practices for production moves

Migrate in staging first, enable wal_compression, monitor pg_stat_progress_copy, and schedule cut-over in low-traffic windows.

Common mistakes and fixes

Missing extensions

If source tables rely on custom types, install the same extensions before running the migration to avoid type errors.

Ignored indexes

Leaving move_indexes=FALSE slows queries.Set it TRUE or recreate needed indexes manually after migration.

Next steps

Create vector indexes with CREATE INDEX ... USING paradedb_ivfflat, update connection strings, and test application queries against the new schema.

FAQ

Is ParadeDB a separate server?

No. ParadeDB runs as a PostgreSQL extension inside the same cluster, so you keep the same connection string.

How long does migration take?

Time depends on data size and I/O. Expect roughly 1–1.3× the time of a plain COPY. Use dry_run to estimate.

.