Q: 9
A company needs to implement a disaster recovery (DR) plan for a web application. The application
runs in a single AWS Region.
The application uses microservices that run in containers. The containers are hosted on AWS Fargate
in Amazon Elastic Container Service (Amazon ECS). The application has an Amazon RDS for MYSQL DB
instance as its data layer and uses Amazon Route 53 for DNS resolution. An Amazon CloudWatch
alarm invokes an
Amazon EventBridge rule if the application experiences a failure.
A solutions architect must design a DR solution to provide application recovery to a separate Region.
The solution must minimize the time that is necessary to recover
from a failure.
Which solution will meet these requirements?
Options
Discussion
Option C looks right. Had something like this in a mock and the cross-Region read replica with ability to promote cuts failover time way down. A and B need to build the DB from snapshot every time, which takes longer. Anyone disagree?
Looks like C is best, but I'm not totally sure. Cross-Region read replica gets you faster recovery than snapshots, since it's already up to date. I've seen a similar Q in practice exams. Agree?
My pick: C, cross-Region read replica gets you the fastest RTO since it's ready to promote instantly. I think AWS expects this warm standby setup if minimizing downtime is the main ask, but open to counterpoints.
My pick: it's C since a cross-Region read replica lets you promote almost instantly for faster RTO. B would add more downtime because you have to spin up the DB from a snapshot, which takes longer. If they wanted lowest cost, B could work but the key here is minimizing recovery time.
C , but I'd double-check with the AWS SA Pro official guide or practice test. Similar exam questions usually push for fastest RTO and mention cross-region replicas.
Option B
C , but I'd check the official guide and some practice exams just to be sure.
C tbh, since the cross-Region read replica lets you promote quickly and have the compute layer hot in the other region. A and B seem slower because you have to spin up the DB from snapshots when disaster happens, which takes more time. D is a bit of a trap, RDS snapshots can't just convert to DynamoDB global tables. I think C is right but open if anyone sees a catch with this setup.
B . Creating resources (ECS cluster, RDS instance from snapshot) only when needed is classic DR, and it does use Lambda for automation. Thing is, if your database is small or downtime isn't super critical, snapshots might be fast enough. I just wonder if cross-Region replication might break if you have heavy DB writes though. Not 100% sure, but that's how I've seen similar setups.
B tbh, because it spins up the resources only when needed and takes a new snapshot, which sounds like typical DR. Though I know it's slower than a read replica approach, the trap is thinking fastest recovery isn't always the main ask.
Be respectful. No spam.
