Architecture and Technical Evaluation: NanoID vs. UUID
Designed by Andrey Sitnik • Standard URL-Safe Alphabet: A-Za-z0-9_-
Information Density & Shannon Entropy Mathematics
Standard UUIDs utilize a 16-character hexadecimal alphabet (0-9, a-f), where each character encodes only 4 bits of entropy ($2^4 = 16$). Standard UUIDs also insert 4 mandatory hyphen delimiters, consuming 36 bytes of storage for 122 bits of randomness (a density of ~3.38 bits/char).
NanoID solves this payload bloat by leveraging a 64-character alphabet (A-Za-z0-9_-), where each character encodes a full 6 bits of entropy ($2^6 = 64$):
Entropy Equation: E = L × log₂(Alphabet Size)
Standard NanoID (L = 21, Alphabet = 64): E = 21 × 6 = 126 bits of entropy
Architectural Comparison: NanoID vs. UUID v4 vs. ULID
| Metric / Property | NanoID | UUID v4 | ULID |
|---|---|---|---|
| Character Length | 21 chars (40% shorter) | 36 chars | 26 chars |
| Total Entropy | 126 bits | 122 bits | 128 bits (48b time + 80b rand) |
| URL / Slug Safety | 100% Native Safe (no escape needed) | Safe but lengthy | 100% Native Safe |
| Alphabet Flexibility | Customizable (e.g. numeric, base36) | Fixed Hex (0-9, a-f) | Fixed Crockford Base32 |
| Ideal Use Case | Short links, frontend routing, NoSQL document keys. | General API tokens & session identifiers. | Time-ordered key-value stores. |
Implementation Snippets
// TypeScript / JavaScript (nanoid package)
import { nanoid, customAlphabet } from 'nanoid';
// 1. Standard 21-character ID:
const id = nanoid();
console.log(id); // "V1StGXR8_Z5jdHi6B-myT"
// 2. Custom Alphabet (e.g. 6-digit numeric OTP / verification pin):
const generatePin = customAlphabet('0123456789', 6);
console.log(generatePin()); // "849201" // Python (nanoid library)
from nanoid import generate
# 1. Standard 21-char secure ID:
nid = generate()
print(f"NanoID: {nid}")
# 2. Custom alphabet & 10-char slug:
custom_slug = generate('1234567890abcdef', 10)
print(f"Slug: {custom_slug}")