OID (Object Identifier)

The OID (Object Identifier) is an internal numeric identifier PostgreSQL uses to refer to system objects: tables, data types, functions, schemas, roles. It’s an unsigned 4-byte integer managed by PostgreSQL itself, distinct from the primary keys of user tables.

How it works #

Every system catalog object (e.g. pg_class for tables, pg_type for types, pg_enum for ENUM values) has an oid column acting as the unique identifier. OIDs are assigned automatically by the engine and used as keys in JOINs across system catalogs. PostgreSQL exposes several conversion functions (oid::regclass, oid::regtype, etc.) to obtain the human-readable name of an object from its OID.

What it’s for #

To identify every database object uniquely and stably across dump-restore. For ENUM types, each value declared in CREATE TYPE ... AS ENUM is assigned an OID, which gets stored in the rows of the tables that use the type. This lets you keep the value in just 4 bytes while still tying it to the readable name and the positional ordering.

When to use it #

Rarely directly in applications — the OID is an implementation detail invisible to most queries. It becomes relevant when analysing system catalogs (information_schema, pg_catalog), when writing introspection or monitoring tooling, and when debugging the behaviour of complex types like ENUMs or domains.