Which architecture would meet these requirements at the LOWEST cost?
Use an Application Load Balancer for Region failover. Deploy new EC2 instances with the userdata script. Deploy separate RDS instances in each Region.
Use Amazon Route 53 for Region failover. Deploy new EC2 instances with the userdata script. Create a read replica of the RDS instance in a backup Region.
Use Amazon API Gateway for the public APIs and Region failover. Deploy new EC2 instances with the userdata script. Create a MySQL read replica of the RDS instance in a backup Region.
Use Amazon Route 53 for Region failover. Deploy new EC2 instances with the userdata script for APIs, and create a snapshot of the RDS instance daily for a backup. Replicate the snapshot to a backup Region.
Explanations:
Using separate RDS instances in each Region would increase cost significantly due to the need for fully provisioned database instances in both Regions. The solution does not address the RPO or RTO effectively.
While using Amazon Route 53 for failover and a read replica helps with some disaster recovery needs, it doesn’t fully meet the RPO of 24 hours because replication lag could exceed this time. The cost may also be higher with a read replica in a separate Region.
While Amazon API Gateway and Route 53 offer cost-effective solutions for public API failover, the MySQL read replica is still an expensive option for disaster recovery. The replication lag may exceed the 24-hour RPO.
This option uses Route 53 for Region failover, EC2 instances for APIs, and replicates RDS snapshots to a backup Region. This solution meets the RTO and RPO requirements at a lower cost by leveraging snapshot replication rather than fully provisioned instances or read replicas.
It appears that the answer is:
Use Amazon Route 53 for Region failover. Deploy new EC2 instances with the userdata script for APIs, and create a snapshot of the RDS instance daily for a backup. Replicate the snapshot to a backup Region.