UUID Generator Guide: v4 vs v7 and When to Use Each

UUIDs enable distributed ID generation

Universally unique identifiers let independent services create IDs without a central allocator. No coordination, no single point of failure, no ID collision under normal conditions. UUIDs appear in API resource IDs, database primary keys, file names, and correlation tokens.

Zovaty UUID Generator

The UUID generator creates v4 (random) and v7 (time-ordered) UUIDs locally in your browser. Generate one or batch multiple. Copy individually or export the list.

UUID v4 vs UUID v7

UUID v4 is purely random. Simple, universally supported, but random insertion order fragments B-tree indexes in PostgreSQL and MySQL.

UUID v7 embeds a Unix timestamp in the high bits. New rows cluster naturally in time-ordered indexes while remaining globally unique. Prefer v7 for database primary keys on high-write tables.

Database indexing tips

Prefer UUID v7 for primary keys on high-write tables

Store as native UUID/binary type, not varchar, when supported

Avoid converting UUIDs to strings in hot query paths

Use v7's embedded timestamp for creation ordering without a separate column

UUIDs in API design

Use UUIDs as public resource identifiers in REST APIs. They are non-sequential, preventing enumeration attacks that incremental integers enable. Include UUIDs in URL paths: /api/users/{uuid}/orders/{uuid}.

Collision probability

For UUID v4, collision probability is negligible at any practical scale. You do not need deduplication logic unless your system has specific requirements. Treat generated UUIDs as unique.

Related Zovaty tools

UUID generator — v4 and v7 generation

JSON formatter — inspect API payloads with UUIDs

Hash generator — complementary identifiers

Conclusion

Use UUID v7 for database keys, v4 for general-purpose random IDs. Generate instantly with the UUID generator on Zovaty Tools.

UUID standards and RFCs

UUID v4 is defined in RFC 4122 using random or pseudo-random data. UUID v7 is defined in RFC 9562 (2024) with timestamp-based ordering. Both are valid for new systems. Avoid UUID v1 (MAC address based) and v3/v5 (namespace hash based) unless you have specific compatibility requirements.

Performance at scale

Generating millions of UUIDs per second is feasible on modern hardware. The bottleneck is database insertion, not generation. UUID v7's time ordering reduces index fragmentation compared to v4 at high insert rates.

Testing with UUIDs

Use deterministic UUIDs in test fixtures for reproducible tests. Generate random UUIDs in integration tests to catch assumptions about ID format. Never hardcode UUIDs from production in test environments.

UUIDs in distributed systems

Microservices generating IDs independently need UUIDs to avoid coordination overhead. Event sourcing systems use UUIDs as event identifiers. Message queues use UUIDs for deduplication keys. The common thread: multiple writers without a central ID allocator.

Storage format considerations

Store UUIDs as native UUID type in PostgreSQL (16 bytes) rather than varchar (36 bytes with hyphens). MySQL supports BINARY(16) storage. Proper typing improves index efficiency and query performance at scale.

Migrating from auto-increment to UUID

Add UUID column alongside existing integer ID. Populate UUIDs for existing rows. Switch application to use UUIDs for new records. Deprecate integer ID after transition period.

UUID alternatives: ULID and Nano ID

ULID provides similar time-ordering to UUID v7 with lexicographic sortability. Nano ID generates shorter URL-friendly IDs. Evaluate based on length requirements, sortability needs, and library support in your stack.

When to use shorter identifiers

Short IDs (Nano ID, CUID) trade collision resistance for brevity and URL friendliness. Use when IDs appear in public URLs and length matters. Use UUIDs when global uniqueness across systems is the priority.

Using UUIDs in test suites

Generate fresh UUIDs for each test run to prevent test interdependence. Use fixed UUIDs only when testing UUID-specific logic like sorting or formatting. The UUID generator provides both random and copyable output.

Summary: UUID usage guide

v7 for database keys. v4 for general random IDs. Store as native type. Use in API public identifiers. Generate with UUID generator.