- What is the difference between ULID and UUID?
- ULIDs are lexicographically sortable by creation time (timestamp prefix), while UUID v4 is fully random. ULIDs are 26 characters with no hyphens; UUIDs are 36 characters with hyphens. Both are 128-bit identifiers.
- Can I use ULID as a database primary key?
- Yes, and it is often better than UUID v4. ULID's time-ordered prefix means inserts are sequential in B-tree indexes, reducing page splits and improving write performance in PostgreSQL and MySQL.
- Are ULIDs unique enough for production?
- Yes. Each millisecond has 80 bits of randomness (2^80 possible values), making collisions practically impossible even in high-throughput distributed systems.
- Can I extract the timestamp from a ULID?
- Yes. The first 10 characters encode a 48-bit millisecond timestamp. Most ULID libraries provide a function to extract the creation timestamp from any ULID.
- Should I replace all UUIDs with ULIDs?
- Consider ULIDs when you need sortable IDs (database keys, event streams). Keep UUIDs for existing systems where changing ID format would require migration, or when you need RFC 4122 compliance.
- Is ULID compatible with UUID columns in databases?
- Yes. ULIDs are 128 bits, the same as UUIDs — so you can store them in a `uuid` column by converting to standard 8-4-4-4-12 hex format. Most ULID libraries provide ulid.toUUID() and ulid.fromUUID() helpers. PostgreSQL: `INSERT INTO users (id) VALUES (ulid_to_uuid('01ARZ3NDEKTSV4RRFFQ69G5FAV'::text));`
- Is ULID an official standard?
- Not an RFC — ULID is a community specification published by alizain/ulid in 2016. It's widely adopted (libraries exist for every major language) but not officially standardized like UUID. If you need a sortable RFC-standard ID, use UUID v7 (RFC 9562, published 2024) — it provides the same time-prefix benefit.
- Can two ULIDs have the same timestamp?
- Yes — ULIDs generated within the same millisecond share the first 10 characters (the timestamp). The last 16 chars (80 bits random) differentiate them. The chance of a random-collision within one millisecond across a single generator is 1 in 2^80 — astronomically improbable. Across distributed generators it's higher but still negligible (need 2^40 ULIDs in one ms for ~50% collision probability — billions per ms).
- What's the maximum ULID value and timestamp range?
- The timestamp portion is 48 bits, covering 2^48 milliseconds = ~8,925 years from 1970 = until the year 10889. The random portion is 80 bits. The maximum ULID is `7ZZZZZZZZZZZZZZZZZZZZZZZZZ`. ULIDs do NOT wrap or overflow within any realistic timeframe.
- Are ULIDs URL-safe?
- Yes — completely. ULIDs use only uppercase A-Z (excluding I/L/O/U) and 0-9. No dashes, no slashes, no special characters. They work directly in URLs, file paths, S3 keys, MongoDB _id fields, and database column values without any escaping.
- Can I extract the creation timestamp from a ULID?
- Yes — that's the whole point of the timestamp prefix. The first 10 chars decode to Unix milliseconds. In JavaScript: `parseInt(ulid.slice(0, 10), 32)`... but use a library function (ulid.decodeTime()) since Crockford Base32 isn't the same as standard Base32 and your parseInt() result will be wrong.
