UUID Generator

How to read a UUID

Take 01a173cb-94ec-7080-a7be-d7d757df7498:

• The first digit of the third group is the version: 4 or 7.
• The first digit of the fourth group is the variant: 8, 9, a or b for standard RFC UUIDs.
• In v7, the first 12 hex digits (01a173cb94ec) are the creation time in milliseconds since 1970. The rest is random, apart from a short counter described below.
• In v4, everything except the version and variant digits is random.

Bulk generation and ordering

When you generate several v7 UUIDs, they are often created within the same millisecond. To keep them in order, the generator uses the 12 bits after the version digit as a counter, starting at a random value, as RFC 9562 suggests: in the example above, 7080, 7081 and 7082. A list of v7 UUIDs is therefore always sorted by creation, both as text and as bytes.

Formats

• Lowercase with hyphens is the canonical form and what most languages and databases print.
• Uppercase matches Microsoft tools and some older systems. UUID comparison is case-insensitive by the standard, but string comparison in code often is not, so pick one form and stick to it.
• Without hyphens, 32 characters, fits fixed-width columns and file names. Store UUIDs in a native uuid or 16-byte binary column where possible instead of text: 16 bytes instead of 36 characters.

Common mistakes

• Using v4 as a primary key in a large table. Random keys scatter inserts across the index and slow down writes. Use v7.
• Exposing v7 IDs where time is sensitive. Anyone can read the creation time from a v7 UUID. Use v4 for public identifiers if that matters.
• Generating IDs with Math.random. It is not designed to be unpredictable or unique. Use crypto.randomUUID() or a proper library in code.