Learn how Neon's autoscaling works - it estimates Postgres' working set size and keeps it in memory. Engineering post here

Serverless

Postgres with instant provisioning, no server management, and pay-per-usage billing

Neon takes the world's most loved database — Postgres — and delivers it as a serverless platform, enabling teams to ship reliable and scalable applications faster.

Enabling serverless Postgres begins with Neon's native decoupling of storage and compute. By separating these components, Neon can dynamically scale up during periods of high activity and down to zero when idle. Developers can be hands-off instead of sizing infrastructure manually.

This serverless character also makes Neon databases highly agile and well-suited for use cases that require automatic creation, management, and deletion of a high number of Postgres databases, like database-per-user architectures with thousands of tenants, as well as database branching workflows that accelerate development by enabling the management of dev/testing databases via CI/CD.

Multi-tenant storage

Read our Architecture section for more information on how Neon is built.

What “serverless” means to us

At Neon, we interpret “serverless” not only as the absence of servers to manage but as a set of principles and features designed to streamline your development process and optimize operational efficiency for your database.

To us, serverless means:

  • Instant provisioning: Neon allows you to spin up Postgres databases in seconds, eliminating the long setup times traditionally associated with database provisioning.
  • No server management: You don’t have to deal with the complexities of provisioning, maintaining, and administering servers. Neon handles it all, so you can focus on your application.
  • Autoscaling: Compute resources automatically scale up or down based on real-time demand, ensuring optimal performance without manual intervention.
  • Usage-based pricing: Your costs are directly tied to the resources your workload consumes—both compute and storage. There's no need to over-provision or pay for idle capacity.
  • Built-in availability and fault tolerance: We’ve designed our architecture for high availability and resilience, ensuring your data is safe and your applications are always accessible.
  • Focus on business logic: With the heavy lifting of infrastructure management handled by Neon, you can dedicate your time and effort to writing code and delivering value to your users.

To us, serverless does not mean…

That Neon only works with serverless architectures. Neon is fully compatible with the entire PostgreSQL ecosystem. Whether you're using Django, Rails, or even a bash script in your basement, if it works with Postgres, it works with Neon.

That you have to pay per query. Your charges are based on compute and storage usage, not the number of queries. For example, you could run billions of queries for as little as $19 a month if they fit within the resources allotted in the Launch plan. The CPU allowance is ample for running sites 24/7 with low CPU requirements.

That you’ll get unpredictable costs due to traffic spikes. We provide transparency in your potential costs. You always set a maximum autoscaling limit to avoid unpredictable bills, and you can always check your consumption. We send you notifications if your storage usage grows quickly.

Learn more

Last updated on

Was this page helpful?