UUID TYPEData Types
Buy Me A Coffee

A native 128-bit column type for storing universally unique identifiers, distinct from storing UUIDs as plain strings.

Globally unique identifiers that can be generated independently across systems without coordination, unlike auto-increment. UUIDs avoid ID conflicts in distributed systems but are larger (16 bytes), slower to index, and harder to debug. Consider UUID v7 for time-ordered IDs.

Compatibility

Show:
Database System Support Status Since Version Notes
MySQL ✗ Not Supported — No native UUID column type. Common options: CHAR(36) for human-readable storage, or BINARY(16) with UUID_TO_BIN() / BIN_TO_UUID() (8.0+) for efficient indexed storage. BINARY(16) is strongly preferred for primary keys.
PostgreSQL ✓ Supported 8.3 Native UUID type stores 16 bytes internally (no hyphen overhead). Displays as the standard 8-4-4-4-12 hyphenated string. Supports equality, indexing, and default expressions with gen_random_uuid().
SQL Server ✓ Supported 2000 UNIQUEIDENTIFIER is the native 16-byte UUID type. Supports NEWID() and NEWSEQUENTIALID() as defaults. Note: SQL Server stores bytes in a mixed-endian order that differs from RFC 4122, which affects cross-platform comparisons.
Oracle ✗ Not Supported — No native UUID column type. SYS_GUID() returns RAW(16), and RAW(16) is the conventional storage type for UUIDs. The standard hyphenated format requires manual conversion.
SQLite ✗ Not Supported — No native UUID type. Store as TEXT (36-char hyphenated string) or BLOB (16 bytes). TEXT is more portable; BLOB is more space-efficient.

Details

PostgreSQL and SQL Server are the only engines with a true native UUID column type that stores 16 bytes internally and handles formatting transparently. MySQL requires a deliberate choice between CHAR(36) (readable, wasteful) and BINARY(16) (efficient but requires conversion functions on read/write). Oracle's RAW(16) is functional but the formatting overhead is significant. A critical SQL Server gotcha: UNIQUEIDENTIFIER byte order is not standard RFC 4122 — the first three components are stored in little-endian order, which causes the displayed string to differ from what other systems would show for the same 16 bytes. This matters when exchanging UUIDs with non-SQL-Server systems. For random (v4) UUIDs as clustered primary keys in SQL Server, prefer NEWSEQUENTIALID() to avoid index fragmentation.

Standard Syntax

-- PostgreSQL CREATE TABLE t ( id UUID PRIMARY KEY DEFAULT gen_random_uuid() );

Version Support

MySQL: Not supported PostgreSQL: Since 8.3 SQL Server: Since 2000 Oracle: Not supported SQLite: Not supported

Per-Database Syntax & Notes

MySQL

No native UUID column type. Common options: CHAR(36) for human-readable storage, or BINARY(16) with UUID_TO_BIN() / BIN_TO_UUID() (8.0+) for efficient indexed storage. BINARY(16) is strongly preferred for primary keys.

-- Efficient storage (8.0+): CREATE TABLE t ( id BINARY(16) DEFAULT (UUID_TO_BIN(UUID(), 1)) PRIMARY KEY ); SELECT BIN_TO_UUID(id) FROM t;

PostgreSQL

Native UUID type stores 16 bytes internally (no hyphen overhead). Displays as the standard 8-4-4-4-12 hyphenated string. Supports equality, indexing, and default expressions with gen_random_uuid().

CREATE TABLE t ( id UUID PRIMARY KEY DEFAULT gen_random_uuid() ); SELECT * FROM t WHERE id = 'a0eebc99-9c0b-4ef8-bb6d-6bb9bd380a11';

SQL Server

UNIQUEIDENTIFIER is the native 16-byte UUID type. Supports NEWID() and NEWSEQUENTIALID() as defaults. Note: SQL Server stores bytes in a mixed-endian order that differs from RFC 4122, which affects cross-platform comparisons.

CREATE TABLE t ( id UNIQUEIDENTIFIER DEFAULT NEWID() PRIMARY KEY ); SELECT * FROM t WHERE id = 'a0eebc99-9c0b-4ef8-bb6d-6bb9bd380a11';

Oracle

No native UUID column type. SYS_GUID() returns RAW(16), and RAW(16) is the conventional storage type for UUIDs. The standard hyphenated format requires manual conversion.

CREATE TABLE t ( id RAW(16) DEFAULT SYS_GUID() PRIMARY KEY ); -- Retrieve as formatted string: SELECT LOWER(REGEXP_REPLACE(RAWTOHEX(id), '([A-F0-9]{8})([A-F0-9]{4})([A-F0-9]{4})([A-F0-9]{4})([A-F0-9]{12})', '\1-\2-\3-\4-\5')) FROM t;

SQLite

No native UUID type. Store as TEXT (36-char hyphenated string) or BLOB (16 bytes). TEXT is more portable; BLOB is more space-efficient.

CREATE TABLE t ( id TEXT PRIMARY KEY ); -- Application generates and inserts the UUID string