ULID vs UUID for database keys?

Use ULID when you need time-ordered records (better index locality, range queries, chronological sorting). Use UUID v4 when you need pure randomness (no timing information leaked). Use UUID v7 (time-ordered, RFC 9562) as a standardized alternative to ULID. ULID is 26 chars; UUID is 36 (with hyphens) or 32 (without).

Does ULID leak timing information?

Yes — the first 10 characters encode the millisecond timestamp. This reveals when the ID was created. For most applications, this is fine (creation time is not secret). For privacy-sensitive applications (anonymous posts, medical records), use a random ID (UUID v4, NanoID) instead. You can also use ULID with a random start time (offset) to obscure the actual time.

Générer ULIDs for time-sortable identifiers

ULID (Universally Unique Lexicographically Sortable Identifier) combines un timestamp avec random données, producing IDs ce trier par creation time.

1

ULID is a 26-character string: 10 chars for timestamp (milliseconds, 48 bits) + 16 chars for randomness (80 bits). ULIDs are lexicographically sortable — sorting ULIDs as strings sorts them by creation time. This is valuable for database indexes (time-ordered), log entries, and event sourcing. ULIDs are URL-safe (Crockford Base32 alphabet: 0-9A-Z, excluding I, L, O, U).
2

Générateur d'UUID / ULID →
Générer a UUID for comparison. UUID v4 is random — sorting UUIDs gives no meaningful order. ULID sorts by time — useful for database primary keys where time-ordered records improve cache locality and range query performance. Use UUID for pure uniqueness; use ULID when time-ordering matters.
3

Convertisseur d'horodatage →
Obtenir the current timestamp to understand the ULID time component. The first 10 characters of a ULID encode the millisecond timestamp. This makes ULIDs decodable — you can extract the creation time without a database lookup. Use this for debugging and log analysis.
4

Encodeur Base62 →
For alternative sortable IDs, encode a timestamp + random number in Base62. This is similar to ULID but uses a larger alphabet (62 chars vs 32), producing shorter IDs. However, ULID's Crockford Base32 is more robust (no ambiguous characters) and standardized.

Questions fréquentes

ULID vs UUID for database keys?: Use ULID when you need time-ordered records (better index locality, range queries, chronological sorting). Use UUID v4 when you need pure randomness (no timing information leaked). Use UUID v7 (time-ordered, RFC 9562) as a standardized alternative to ULID. ULID is 26 chars; UUID is 36 (with hyphens) or 32 (without).
Does ULID leak timing information?: Yes — the first 10 characters encode the millisecond timestamp. This reveals when the ID was created. For most applications, this is fine (creation time is not secret). For privacy-sensitive applications (anonymous posts, medical records), use a random ID (UUID v4, NanoID) instead. You can also use ULID with a random start time (offset) to obscure the actual time.