What is ULID?

ULID (Universally Unique Lexicographically Sortable Identifier) is a 128-bit identifier encoded as a 26-character string. Unlike UUID v4, ULIDs sort in creation order, making them excellent primary keys for databases and distributed logs.

Structure

The first 10 characters encode the Unix timestamp in milliseconds using Crockford Base32 (highlighted in amber). The last 16 characters are 80 bits of cryptographically random data, generated via crypto.getRandomValues().

ULID vs. UUID comparison

Property	ULID	UUID v4	UUID v7
Sortable by creation time	Yes	No	Yes
Timestamp embedded	Yes (ms)	No	Yes (ms)
Encoded length	26 chars	36 chars	36 chars
Character set	Crockford Base32	Hex + hyphens	Hex + hyphens
Collision probability (80-bit random)	Very low	Very low (122-bit)	Very low (74-bit)
B-tree index friendly	Yes	No	Yes

UUID v7 is a close counterpart to ULID: both embed a millisecond Unix timestamp for sortability. The key difference is encoding - ULID’s Crockford Base32 is URL-safe and case-insensitive, while UUID v7 uses the familiar hex-hyphen format with wider library support.

Database primary key guidance

Because ULIDs sort lexicographically by creation time, new records are always inserted near the end of a B-tree index. This dramatically reduces index page splits and fragmentation compared to random UUIDs, which insert at random positions and force costly page reorganization. In high-write workloads on PostgreSQL or MySQL, switching from UUID v4 to ULID (or UUID v7) can cut index write amplification significantly and improve insert throughput.

Crockford Base32 alphabet

ULID uses Douglas Crockford’s Base32 alphabet, which deliberately excludes four characters: I, L, O, and U. These are omitted because they are visually ambiguous with 1, 1, 0, and V respectively, making handwritten or spoken ULIDs much less error-prone. The resulting alphabet is 0123456789ABCDEFGHJKMNPQRSTVWXYZ.