Unique identifiers: Differences between UUID v4, v7 and ULID in databases

When designing distributed computing systems, assigning primary keys in databases requires identifiers that never collide. For years, the standard solution has been UUID v4. However, modern computing is adopting more efficient alternatives such as UUID v7 and ULID.

The evolution of 128-bit IDs

• UUID v4 (Random): Excellent for ephemeral tokens. However, its complete randomness destroys the performance of the primary index in relational databases when inserting records at a large scale.
• UUID v7 (Ordered Time): The new official standard (RFC 9562). It combines 48 bits of timestamp with random bits, meaning IDs are inserted sequentially dramatically improving read/write performance.
• ULID (Universally Unique Lexicographically Sortable Identifier): Similar to UUID v7, but uses Base32 encoding (26 characters) making it more compact to store and represent in URLs than traditional UUID scripts.

To generate UUID v4, v7 or ULID identifiers in bulk and locally for your software developments, use our interactive tool:

Try our UUID and ULID Generator

Instantly generate random or chronological IDs ready to use in your migration scripts or databases.