Neon Serverless Postgres for Reliable and Scalable Applications

What is Neon?

Neon offers a fully managed, serverless PostgreSQL database platform engineered for modern application development. It enables developers to focus on building applications by automating infrastructure management tasks. Key capabilities include instant database provisioning, automatic scaling based on workload demands (including scale-to-zero functionality to optimize costs), and efficient connection pooling to handle thousands of concurrent connections.

The platform incorporates innovative features such as database branching, allowing developers to instantly create isolated copies of their database for development, testing, CI/CD pipelines, and schema migrations using copy-on-write storage. This significantly accelerates development workflows. Neon also supports instant point-in-time recovery (PITR) and provides robust management options through its API and CLI. Additionally, it is optimized for AI applications, supporting vector search functionalities through extensions like pgvector, making it suitable for building cutting-edge AI-powered features.

Features

Autoscaling: Automatic instance sizing and scale-to-zero based on load.
Branching: Instantly create isolated database copies for development and testing using copy-on-write.
Connection Pooler: Handles thousands of database connections efficiently.
Instant Point-in-Time Recovery (PITR): Restore data up to 30 days granularity.
Serverless Postgres: Fully managed PostgreSQL without server management.
API & CLI Management: Programmatically manage databases.
PGVECTOR Support: Enables efficient high-dimensional vector search for AI applications.

Use Cases

Developing and scaling SaaS applications.
Building serverless applications requiring scalable databases.
Implementing database-per-tenant architectures efficiently.
Creating production-like environments for development and testing.
Deploying PostgreSQL databases for AI agents and applications.
Accelerating development cycles with database branching.
Powering AI applications requiring vector search capabilities.

FAQs

What is a compute hour in Neon?

A compute hour is the metric for compute usage, calculated as [compute size] x [hours your compute runs]. Each pricing plan includes a set number of compute hours, with overages billed monthly.
How does billing work in Neon?

You are charged a monthly fee based on your plan, plus any additional compute and/or storage usage beyond the plan's included amounts. Billing occurs at the beginning of each calendar month for the previous month's usage.
How can I control my costs with Neon?

By setting a maximum autoscaling limit (e.g., 2 CU) when enabling autoscaling, you establish a de-facto cost limit for compute usage.
How is storage charged in Neon?

Neon uses a unique storage engine with copy-on-write, meaning creating database branches does not duplicate storage or add significantly to the storage bill.
Is Neon suitable for database-per-user architectures?

Yes, Neon recommends a project-per-user (or project-per-tenant) pattern, leveraging the high project limits in its pricing plans for such architectures.