UUID vs ULID Comparison

ULID (Universally Unique Lexicographically Sortable Identifier) was created in 2016 as an alternative to UUID that prioritizes sortability and human readability. With the introduction of UUID v7, the two formats now overlap significantly in functionality.

Format comparison:

Property	UUID	ULID
Size	128 bits	128 bits
String length	36 chars (with hyphens)	26 chars (Crockford Base32)
Sortable	v7 only	Always
Timestamp	v7: 48-bit ms	48-bit ms
Randomness	v4: 122 bits, v7: 62 bits	80 bits
Encoding	Hexadecimal	Crockford Base32

ULID structure (128 bits):

 01AN4Z07BY      79KA1307SR9X4MV3
|----------|    |----------------|
 Timestamp          Randomness
  48 bits            80 bits
  (10 chars)         (16 chars)

A ULID looks like 01ARZ3NDEKTSV4RRFFQ69G5FAV — no hyphens, uppercase, 26 characters. The Crockford Base32 encoding excludes I, L, O, and U to avoid ambiguity with 1, l, 0, and obscene words.

UUID v7 vs ULID — key differences:

1. Encoding: ULIDs use Base32 (26 chars) vs UUID's hex (36 chars), saving 10 characters per ID. This adds up in URLs, logs, and APIs.
2. Randomness: ULID has 80 random bits vs UUID v7's 62 bits, giving ULID a lower collision probability within the same millisecond.
3. Ecosystem: UUID is an IETF standard (RFC 9562) with native support in virtually every database, language, and framework. ULID is a community specification with growing but less universal support.
4. Compatibility: ULIDs can be stored in UUID columns (both are 128 bits), but the string representation is different, which can cause confusion.

When to choose ULID: Prefer ULID when string length matters (URLs, user-facing IDs), you do not need database-native UUID support, and your ecosystem already uses ULID libraries.

When to choose UUID v7: Prefer UUID v7 when you need broad compatibility with databases (PostgreSQL uuid type, MySQL BINARY(16)), standards compliance matters, and your tools already understand UUID format.

Performance: Both perform equally well for database indexing because both are time-ordered. The storage cost is identical at 128 bits (16 bytes) in binary. The only difference is string representation size.