What does redundancy mean, and why does it matter?

Your app runs on several servers in different countries, so your users see no interruption even if a whole region goes down.

The life of an application running on a single server is tied to the life of that server. Hardware fails, a data centre loses power, a network problem makes a region unreachable. These are not rare; only their timing is unknown.

What is a copy?

Redundancy means your app runs on more than one server at the same time. We call each one a copy. Running with two copies, your app stands on two different servers, often in different countries, and incoming requests are distributed between them.

What happens when one goes down? The health check notices, takes that copy out of rotation and routes traffic to the one that is up. Your visitor feels nothing. Meanwhile a replacement for the failed copy is deployed.

How many copies are enough?

A single copy is the cheapest and is reasonable for personal projects and test environments — but you will see downtime during maintenance or a failure. Two copies are the right balance for most production apps. Three copies are for businesses that find downtime unacceptable.

You can change the number of copies later; raising it for a campaign period and lowering it afterwards is a common pattern.

Redundancy is not a backup

A point that is often confused: redundancy and backups are not the same thing. Redundancy protects against a failure in the moment — if one server goes down the other carries on. But if you delete a table by accident, that deletion is reflected across every copy at once.

If you want to go back in time, you need a backup. Neither replaces the other; they are used together.

Other posts