ULID

Shape

Background

Origin

ULID stands for Universally Unique Lexicographically Sortable Identifier, a format set out in the ulid/spec repository on GitHub.^[1] The ULID specification notes that UUID v4 provides no other information than randomness, which can cause fragmentation in many data structures.^[1] The ULID specification describes itself as the current specification of ULID as implemented in the ulid/javascript library.^[1] The ulid/javascript package names Alizain Feerasta as its author.^[2] The ULID specification says the format was partly inspired by posts on Instagram's engineering blog and Firebase's blog.^[1] RFC 9562 lists ULID first among the 16 implementations analysed while its specification was being prepared.^[3]

A ULID is made of a 48-bit timestamp followed by 80 bits of randomness.^[1] The ULID timestamp is UNIX time in milliseconds.^[1] The ULID specification says the timestamp won't run out of space until the year 10889 AD.^[1] In the canonical ULID string, the first 10 characters encode the timestamp and the remaining 16 encode the randomness.^[1] ULIDs are canonically encoded as a 26-character string, as opposed to the 36-character UUID.^[1][4] ULID uses Crockford's Base32, at 5 bits per character.^[1] Crockford's Base32 alphabet leaves out the letters I, L, O and U.^[1][5] Douglas Crockford's Base 32 page says I and L are left out because each can be confused with 1, and O because it can be confused with 0.^[5] Douglas Crockford's Base 32 page gives accidental obscenity as the reason for leaving out U.^[5] In binary, a ULID is 16 octets, with each component encoded most significant byte first.^[1]

Use

The ULID specification sorts ULIDs in lexical order, left-most character first, using the default ASCII character set.^[1] The ULID specification says that within the same millisecond, sort order is not guaranteed.^[1] In the ULID specification's monotonic mode, when the same millisecond is detected the random component is incremented by 1 bit in the least significant bit position.^[1] Sort on a ULID column for a rough creation order, but do not lean on it to order rows that may share a millisecond, and treat the decoded time as a hint set by whichever system minted the ID rather than as an audit timestamp. (Meridian's view)

Laravel's Eloquent ORM can give a model ULID primary keys through its HasUlids trait.^[6] Laravel's documentation describes ULIDs as lexicographically sortable for efficient database indexing.^[6] Symfony's documentation says using ULIDs as primary keys is usually not recommended for performance reasons.^[4] Keep ULIDs as text when you load them, and if a ULID column will not join against another table, check whether the other side holds the same identifiers in a different form, such as a UUID-style string or raw bytes. (Meridian's view)

Sources

1. [1]GitHub - ulid/spec: The canonical spec for ulid · GitHubdocumentation
2. [2]javascript/package.json at master · ulid/javascript · GitHubdocumentation
3. [3]RFC 9562: Universally Unique IDentifiers (UUIDs)standard
4. [4]The UID Component (Symfony Docs)documentation
5. [5]Base 32secondary source
6. [6]Eloquent: Getting Started | Laravel 12.x - The clean stack for Artisans and agentsdocumentation

Detection notes

Use it

Schema