Post image

We all have experienced what happens when a production database goes down—it’s not fun. No matter the cause, the result is always the same—high stress, downtime, and a rush to restore everything as quickly as possible.

How long would it take to get your Postgres database back online? For most, the answer hours, certainly if your DB has a certain size (TBs). 

What if restoring your database was instant? That’s exactly what we’re about to show you. In this post, we’re going to simulate a database disaster and bring it back to life instantly using Neon’s Instant Restore. Instead of just telling you about it, we built a hands-on Outage Simulator so you can see it in action:

http://fyi.neon.tech/recovery

Try it Yourself: The Outage Simulator 

For this interactive demo, 

  1. We built a simple web app, a social media feed. It loads posts from a real Postgres database running on Neon. 
  2. Then, we simulate a database failure. The app crashes, everything breaks.
  3. To fix it, we trigger an instant restore in Neon. 
  4. The app then comes back online—fast. 

 Watch it happen or run it yourself:

How is This Possible? 

What’s actually happening under the hood?

Traditional Postgres restores involve restoring from snapshots or backups and replaying WAL, a process that takes time, especially for large datasets. Neon is able to take a completely different approach to recovery, leveraging its copy-on-write branching to make restores near-instant.

In the Outage Simulator, this is what we’re doing:

  1. We loaded 1 TB (1254 GB to be exact) into a main branch in a Neon project. 
  2. Instead of having everyone connect to a shared database, each user session gets a dedicated Neon branch. 
  3. When the app fails, we don’t reload a backup. Instead, Neon creates a new branch from the exact point in time before the failure and seamlessly redirects the app to this restored branch. This happens in less than a second.

Thanks to Neon’s architecture, branches don’t copy data; they reference shared storage and only record changes. That’s what makes recovery so fast, even for multi-terabyte databases. If this demo was 100 TB, recovery would be just as quick. 

A New Standard for Database Recovery

Accidents are inevitable. The real question isn’t if they’ll happen, but how quickly you can recover.

Database recovery mechanisms in services like Amazon RDS are slow and complex to maintain, especially at scale. More often than not, restoring takes hours, and every second of downtime means lost revenue, frustrated users, and operational headaches.

Neon changes this. With Instant Restore, you can rewind your database immediately and keep your applications running. Create a Neon account and try this in our Free Plan.