UUID, Nanoid, ULID: Choosing the Right ID Format

The Three Contenders

UUID v4

The most widely used identifier. 128 bits of randomness, formatted as 550e8400-e29b-41d4-a716-446655440000. Universally supported across databases, languages, and frameworks.

Pros: ubiquitous, zero dependencies
Cons: not sortable, verbose (36 chars with hyphens)

Nanoid

A URL-safe 21-character ID using a 64-character alphabet by default. V1StGXR8_Z5jdHi6B-myT

Pros: shorter, URL-safe, customisable alphabet and length
Cons: less universal support, no time ordering

ULID

A 26-character lexicographically sortable ID. The first 10 characters encode a millisecond timestamp, the remaining 16 are random: 01ARZ3NDEKTSV4RRFFQ69G5FAV

Pros: sortable by creation time, monotonic within the same millisecond
Cons: leaks creation timestamp, fewer libraries

Generating via API

curl "https://api.toolkitapi.io/v1/devtools/generate-id?type=ulid&count=5" \
  -H "X-API-Key: $API_KEY"

{
  "ids": [
    "01JQXYZ1234567890ABCDEFGH",
    "01JQXYZ1234567890ABCDEFI",
    ...
  ]
}

Decision Matrix

Database Considerations

Random UUIDs fragment B-tree indexes over time, causing write amplification. ULIDs inserted in time order keep indexes largely sequential, improving insert throughput significantly on large tables.