Neon is a serverless Postgres platform with instant branching, scale-to-zero compute, and a WebSocket-based driver for edge environments. Its separation of compute and storage enables unique workflows like per-PR preview databases.
0 / 5 completed
1 / 5
What is database branching in Neon Serverless Postgres?
Neon branching uses copy-on-write storage to create instant database clones. A branch shares storage pages with its parent until data diverges, making it nearly free to create. This enables workflows like per-PR preview databases or safe schema migration testing without provisioning new servers.
2 / 5
What is the purpose of Neon's serverless driver (@neondatabase/serverless)?
The Neon serverless driver opens connections over WebSockets (or HTTP for single queries), making it compatible with Cloudflare Workers, Vercel Edge Functions, and other environments where raw TCP is unavailable. It is API-compatible with node-postgres (pg) for easy migration.
3 / 5
How does Neon's autoscaling differ from traditional Postgres managed services?
Neon autoscaling separates compute from storage. The compute layer can scale up CPU/RAM within seconds during traffic spikes and scale down to zero after a configurable idle period. Because storage is shared via Neon's distributed storage engine, there is no data migration when scaling compute.
4 / 5
What is connection pooling in Neon, and which protocol does it use?
Neon's connection pooler is a managed PgBouncer instance running in transaction mode. You connect to the pooler endpoint (e.g., ep-xxx-pooler.neon.tech) to multiplex many short-lived serverless connections across a smaller number of actual Postgres connections, preventing the connection exhaustion common in serverless deployments.
5 / 5
What does the point-in-time restore feature in Neon allow you to do?
Point-in-time restore in Neon leverages its branching infrastructure. Because Neon retains WAL history within the retention window, you can branch from any timestamp in that window, effectively cloning the database as it existed at that moment. This is useful for disaster recovery and investigating data at a past state.