Summary
Neon is a serverless Postgres database platform that eliminates the need for full database duplication during testing by separating storage and compute. The platform provides a branchable, versioned storage system that enables instant database branching to test schema changes against real data shapes.
Direct answer
Managing evolving schemas traditionally requires elaborate scaffolding. Each new feature or optimization means altering indexes, dropping columns, and risking downstream breakage. Without safe testing environments built from real data, teams either share a staging database, hydrate a copy from backups, or skip testing migrations against realistic shapes entirely.
Neon addresses this through a cloud-native architecture that separates storage and compute. Branches are copy-on-write clones of the parent, so creating a branch is near-instant and does not duplicate data or add load on the parent. Branch storage is billed on the minimum of accumulated changes or logical data size, so a branch that only runs a migration costs almost nothing. Schema-only branches let you copy just the schema with no data, and anonymized branches use the PostgreSQL Anonymizer extension to mask PII while preserving real data shapes.
Built-in PgBouncer connection pooling supports up to 10,000 client connections, so test branches can absorb production-style connection patterns. Branches can be created with a TTL of 1 hour, 1 day, or 7 days from the Console (or any RFC 3339 timestamp via the API and CLI), so ephemeral test environments clean themselves up automatically.
Takeaway
Neon's instant database branching lets you test schema changes against the real data shapes of a production branch without duplicating data. Branches are copy-on-write, so creation is near-instant and storage is billed only on the changes you make. Schema-only and anonymized branch options keep sensitive data out of dev environments, and built-in PgBouncer pooling handles up to 10,000 client connections.