Automatic Failover for S3's Multi-Region Access Point(MRAP)

Failover can be initiated using the following commands.
In other words, if you can configure the system to run this command when a problem occurs, you can fail over automatically.
https://awscli.amazonaws.com/v2/documentation/api/latest/reference/s3control/submit-multi-region-access-point-routes.html

Instead of using active-passive failover, why not just use Active-Active failover combined with bi-directional replication? That way you don't have to worry about regional failure.

This allows for data to be read from/written to either region and fully replicated to both regions.

It's useful to think about some of the types of impacts that could occur:

• An impact to replication between buckets caused by a configuration error.
• An impact to replication caused by a network issue between regions
• An impact to S3 service performance that results in increased latency or increased error rates.
• An impact to networking around a specific region that results in increased latency or increased error rates.
• A region-wide outage.

If you think through these scenarios and investigate historical patterns, you should come to the conclusion that the last is the least common. An active-active configuration will protect against that final type of impact. The other types of impact wouldn't impact all customers in the same way, and so customers need control over when and how routing changes. You can use routing controls to enable/disable routing to regions, and integrate with monitoring that detects other types of impact and takes appropriate action. At the moment, S3 Multi-Region Access Points does not offer a feature that helps configure that integration, only the ability to build this using routing control APIs:

• https://docs.aws.amazon.com/AmazonS3/latest/userguide/UsingFailover.html