SQL, Postgres OIDs, What are they and why are they useful?

Question

OIDs basically give you a built-in id for every row, contained in a system column (as opposed to a user-space column). That's handy for tables where you don't have a primary key, have duplicate rows, etc. For example, if you have a table with two identical rows, and you want to delete the oldest of the two, you could do that using the oid column.

OIDs are implemented using 4-byte unsigned integers. They are not unique–OID counter will wrap around at 2³²-1. OID are also used to identify data types (see /usr/include/postgresql/server/catalog/pg_type_d.h).

In my experience, the feature is generally unused in most postgres-backed applications (probably in part because they're non-standard), and their use is essentially deprecated:

In PostgreSQL 8.1 default_with_oids is off by default; in prior versions of PostgreSQL, it was on by default.

The use of OIDs in user tables is considered deprecated, so most installations should leave this variable disabled. Applications that require OIDs for a particular table should specify WITH OIDS when creating the table. This variable can be enabled for compatibility with old applications that do not follow this behavior.

OID's are still in use for Postgres with large objects (though some people would argue large objects are not generally useful anyway). They are also used extensively by system tables. They are used for instance by TOAST which stores larger than 8KB BYTEA's (etc.) off to a separate storage area (transparently) which is used by default by all tables. Their direct use associated with "normal" user tables is basically deprecated.

The oid type is currently implemented as an unsigned four-byte integer. Therefore, it is not large enough to provide database-wide uniqueness in large databases, or even in large individual tables. So, using a user-created table's OID column as a primary key is discouraged. OIDs are best used only for references to system tables.

Apparently the OID sequence "does" wrap if it exceeds 4B 6. So in essence it's a global counter that can wrap. If it does wrap, some slowdown may start occurring when it's used and "searched" for unique values, etc.

See also https://wiki.postgresql.org/wiki/FAQ#What_is_an_OID.3F

OIDs being phased out

The core team responsible for Postgres is gradually phasing out OIDs.

Postgres 12 removes special behavior of OID columns

The use of OID as an optional system column on your tables is now removed from Postgres 12. You can no longer use:

• CREATE TABLE … WITH OIDS command
• default_with_oids (boolean) compatibility setting

The data type OID remains in Postgres 12. You can explicitly create a column of the type OID.

After migrating to Postgres 12, any optionally-defined system column oid will no longer be invisible by default. Performing a SELECT * will now include this column. Note that this extra “surprise” column may break naïvely written SQL code.

To remove all OIDs from your database tables, you can use this Linux script:

First, login as PostgreSQL superuser:

sudo su postgres

Now run this script, changing YOUR_DATABASE_NAME with you database name:

for tbl in `psql -qAt -c "select schemaname || '.' || tablename from pg_tables WHERE schemaname <> 'pg_catalog' AND schemaname <> 'information_schema';" YOUR_DATABASE_NAME` ; do  psql -c "alter table $tbl SET WITHOUT OIDS" YOUR_DATABASE_NAME ; done

I used this script to remove all my OIDs, since Npgsql 3.0 doesn't work with this, and it isn't important to PostgreSQL anymore.