In the previous subchapter, we saw disaster recovery strategies where, if the main system fails, traffic must switch to a backup system. But how do you detect that the main one has failed and redirect people automatically, without a human having to intervene at 3 a.m.? The answer combines two functions of Route 53 (the AWS DNS we saw in subchapter 16.1): health checks and failover.

Remember from subchapter 16.1 that Route 53 is AWS's DNS service: it translates a domain name (like myshop.com) to the address of the server that should handle it. It's the first thing a user's browser checks to know where to connect. This gives it a privileged position: Route 53 decides where the traffic goes. And that's the key to automatic failover.

Imagine you have your main system in one region and a backup in another (as in the strategies from subchapter 26.2). If the main one goes down, you need users to stop going to the main (down) system and go to the backup (healthy) one. And you need this to happen:

• Automatically (without waiting for a human to notice and act).
• Quickly (every minute of downtime counts).
• In a reliable way (without sending people to a broken system).

For this, you first need to detect that the main one failed, and then redirect. Route 53 does both.

A Route 53 health check is an automatic monitor that periodically checks if your system is responding correctly. Route 53 "asks" your system every so often: "are you okay?", and based on the response, marks it as healthy or unhealthy.

Analogy: a health check is like taking a patient's pulse every few minutes. As long as the pulse is normal, all is well. If the pulse stops or becomes abnormal, the alarm goes off and action is taken. Route 53 "takes the pulse" of your systems continuously to know which ones are alive and healthy.

The health check can check things like: does the website respond? Does it return a correct code? Does it respond in time? You define what "healthy" means.

Here's the magic. Failover is Route 53's ability to automatically redirect traffic from the main system to the backup when the health check detects that the main one is unhealthy.

Remember the routing policies from subchapter 16.1: one of them is precisely failover. You configure Route 53 like this:

Analogy: failover is like an emergency power generator in a hospital. As long as there's power from the grid (main system healthy), everything works normally. The instant the power goes out (main fails), a system automatically detects the outage and starts the generator (backup) in seconds, without anyone having to run and do it. The hospital keeps running without patients noticing. Health check = outage detector; failover = automatic generator startup.

The two are inseparable: the health check detects, the failover reacts:

Without the health check, Route 53 wouldn't know something failed. Without failover, knowing it failed would be useless. Together, they achieve automatic traffic recovery, which is exactly what makes DR strategies (26.2) work without human intervention.

Real-world example: a company has its main website in the Ireland region and a backup (warm standby, subchapter 26.2) in Frankfurt, with Route 53 configured for failover. One night, the Ireland region has a problem and the website stops responding. The Route 53 health check detects it in seconds and marks Ireland as unhealthy. The failover automatically redirects all users to Frankfurt, which was ready. Customers barely notice a brief interruption. No one on the team had to wake up or do anything: the system recovered by itself. The next morning, when Ireland is restored, traffic automatically returns. That's resilience done right.

These same capabilities (health checks + Route 53 routing policies) are also used to distribute users between regions by proximity (remember the geolocation and latency policies from subchapter 16.1), sending each user to the closest and healthy region. This way, system health is considered not only for emergencies, but also to provide the best service day to day.

• Route 53 (the AWS DNS, subchapter 16.1) decides where traffic goes, allowing it to manage automatic failover.
• A health check periodically checks if a system responds well and marks it as healthy or unhealthy. Like taking a patient's pulse continuously.
• Failover automatically redirects traffic from the main system to the backup when the health check detects the main is unhealthy (and returns it when it recovers). Like an emergency generator that starts by itself when the power goes out.
• They work together: the health check detects, the failover reacts. Together they achieve automatic traffic recovery, without human intervention, making DR strategies work.
• The same capabilities are used for geographic load balancing (sending each user to the closest and healthy region), not just for emergencies.

In the last subchapter of the chapter (and of Part VI) we'll see how to protect your data with centralized and automatic backups: AWS Backup.