Which combination of steps will result in a highly available architecture that meets all the requirements?
(Choose two.)
Create a cross-Region read replica of the database in the secondary Region. Configure an AWS Lambda function in the secondary Region to promote the read replica during failover event.
In the primary Region, create a health check on the database that will invoke an AWS Lambda function when a failure is detected. Program the Lambda function to recreate the database from the latest database snapshot in the secondary Region and update the Route 53 host records for the database.
Create an AWS Lambda function to copy the latest automated backup to the secondary Region every 2 hours.
Create a failover routing policy in Route 53 for the database DNS record. Set the primary and secondary endpoints to the endpoints in each Region.
Create a hot standby database in the secondary Region. Use an AWS Lambda function to restore the secondary database to the latest RDS automatic backup in the event that the primary database fails.
Explanations:
Creating a cross-Region read replica allows for automated synchronization of data between Regions. The Lambda function can be used to promote the read replica to the primary in the event of a failure, providing a fast recovery solution.
While using Lambda to invoke actions during failure is useful, recreating the database from a snapshot every time could cause significant downtime, and it doesn’t meet the RTO of 15 minutes, which requires a more seamless failover solution.
Copying the latest automated backup every 2 hours is a good practice for backup redundancy but does not address the requirement for a highly available, immediately available failover solution within the RTO and RPO.
A failover routing policy in Route 53 allows automatic switching between Regions based on health checks, ensuring that the database traffic is routed to the secondary Region if the primary Region becomes unavailable. This meets the requirements for high availability and automatic failover.
A hot standby database involves maintaining a fully synchronized replica, which would be a better solution than copying backups. However, restoring a backup via Lambda in the event of a failure would not meet the RTO of 15 minutes, as restoring from a backup is slower than promoting a read replica.