Multi-AZ and multi-Region high availability: DOP-C02 study notes and practice questions
What to know about Multi-AZ and multi-Region high availability for the DOP-C02 exam, from the Resilient Cloud Solutions domain (15% of the exam guide), then 3 free questions to test it. Part of the DOP-C02 study guide.
These questions give availability and data loss targets and ask which data and compute setup meets them with least effort.
Within a Region (Multi-AZ)
Service
Option
Behavior
RDS
Multi-AZ DB instance
Synchronous standby in another AZ; automatic failover under the same endpoint; standby doesn’t serve reads
RDS
Multi-AZ DB cluster
Two readable standbys; faster failover
Aurora
Replicas across AZs
Storage is replicated six ways across three AZs; a replica is promoted on failure
ElastiCache
Multi-AZ with automatic failover
Replica promoted
EFS
Regional (standard) file systems
Data stored across AZs
Compute
Auto Scaling group across AZs behind a load balancer
Lose an AZ, keep serving
Read replicas scale reads; they aren’t automatic HA for RDS (manual promotion).
Across Regions
Aurora Global Database: storage-based replication to secondary Regions with lag typically under a second; switchover for planned moves (no data loss) and failover for outages.
DynamoDB global tables:multi-active; every Region accepts writes, replicated to the others (last writer wins by default; multi-Region strong consistency is an option).
S3 Cross-Region Replication with Replication Time Control (99.99% of objects within 15 minutes, with metrics).
RDS cross-Region read replicas and automated backup replication.
ECR, AMIs, secrets and KMS multi-Region keys must be in the other Region too.
Designing apps
Keep the app tier stateless; store sessions in DynamoDB or ElastiCache.
Use Route 53 or Global Accelerator to steer users.
Automate failover runbooks and test them.
Common traps
Read replicas when the ask is automatic failover with no app change. Use Multi-AZ.
Cross-Region read replicas when users in several Regions must write. Use DynamoDB global tables or another multi-active design.
Forgetting AMIs, images and secrets in the recovery Region.
A regional bank runs its core ledger on an Amazon Aurora PostgreSQL cluster in us-east-1, with a writer and two readers spread across three Availability Zones. The cluster already survives the loss of a single Availability Zone without manual work.
A new regulator requires a second Region that can take over the ledger after a regional outage, with a recovery point objective of about one second. The bank also needs to rehearse the failover in production-like conditions every quarter, without losing committed transactions.
Application servers must keep writing through one database endpoint per Region, and the bank wants minimal changes to the application code.
Which approach meets these requirements?
Choose one answer.
AA. Copy the cluster's snapshots to the second Region every hour with AWS Backup, and after a regional outage restore a new cluster from the latest copy and point the application at it.
BB. Enable Multi-AZ for the cluster in a second Region, so that the standby instances in that Region receive synchronous replication and take over automatically during a regional outage.
CC. Add a secondary Region to an Aurora global database, use a global database switchover for the quarterly rehearsals, and fail over to the secondary Region during an outage.
DD. Replace the ledger with an Amazon DynamoDB global table, and point each Region's application servers at the local replica, so that the table replicates writes across Regions automatically.
C. Right: Correct. An Aurora global database replicates to the secondary Region with latency typically under a second, and a switchover moves the primary to that Region without data loss, which suits rehearsals. Applications keep using the cluster endpoints in each Region. Learn more: Using Amazon Aurora Global Database (opens in a new tab)
An Aurora global database extends the cluster to a secondary Region with replication lag typically under a second, so a regional outage can be recovered with a cross-Region failover. A switchover, formerly called managed planned failover, lets the bank rehearse without losing committed data. Snapshot copies, Multi-AZ and DynamoDB global tables each miss the recovery point, the Region or the minimal-change requirement.
A switchover moves the writer with no data loss for planned rehearsals, and a failover promotes the secondary during an outage.
A hospital runs its patient-record reporting database on Amazon RDS for PostgreSQL in one Region, with the application servers in the same VPC. The database currently has a single instance in one Availability Zone.
If the Availability Zone fails, the hospital needs the database to fail over automatically within minutes, with no loss of committed transactions. The clinical applications connect with the database's existing endpoint, and the team cannot change their connection strings.
The hospital does not need cross-Region recovery for this database, and it wants the lowest operational effort.
Which change meets these requirements?
Choose one answer.
AA. Convert the instance to a Multi-AZ DB instance deployment, with a synchronous standby in another Availability Zone and automatic failover under the same endpoint.
BB. Create a read replica in another Availability Zone, and have the team change the connection string to the replica's endpoint whenever the primary becomes unavailable.
CC. Create a cross-Region read replica in another Region, and promote it manually after an Availability Zone failure so that the application gets a new database endpoint.
DD. Take a snapshot every hour, and restore a new instance from the most recent snapshot in another Availability Zone after a failure, then update the application connection strings.
D. Wrong: Restoring from a snapshot takes a long time and loses everything since the last snapshot. It also requires connection string changes, which the team cannot make. Learn more: What is AWS Backup? (opens in a new tab)
Explanation
A Multi-AZ DB instance deployment uses a synchronous standby in another Availability Zone and fails over automatically under the same endpoint, so the hospital needs no application changes. Read replicas, cross-Region replicas and snapshot restores each require endpoint changes or manual work, and asynchronous copies can lose recent transactions.
A university runs a student-services application that must stay usable for students in two countries if either Region becomes unavailable. Students write to their own profile records from the Region nearest to them, and both Regions must eventually hold the same data.
Uploaded student documents are stored in Amazon S3 in the primary Region and must be copied to a second Region within a short time, with replication time control enabled so that copies are predictable.
The university wants the solution to use managed AWS features, and it does not want to run its own synchronization service.
Which TWO actions meet these requirements? (Choose two.)
Choose two answers.
AA. Keep the application in a single Region with Multi-AZ deployments for the database, and add Route 53 failover to a standby web tier in the second Region that serves read-only pages.
BB. Create a DynamoDB global table with replicas in both Regions for the profile records, so that writes in each Region replicate to the other with no custom synchronization.
CC. Use a nightly S3 Batch Operations job that copies the documents from the primary Region to the second Region, so the copy happens on a schedule that the operations team controls.
DD. Turn on S3 Cross-Region Replication for the document bucket with replication time control enabled, so that new objects are copied to the second Region within the service's replication time target.
EE. Store profile records in Amazon Aurora with a single writer in the primary Region, and let students in the second Region reach that writer over a VPC peering connection.
E. Wrong: Students in the second Region would still depend on the primary Region for writes, so a regional outage would stop their updates. The design does not meet the requirement that both Regions can accept writes. Learn more: What is VPC peering? (opens in a new tab)
Explanation
DynamoDB global tables let each Region accept writes that replicate to the others, which suits profile records that students update locally. S3 Cross-Region Replication with replication time control copies new documents within a predictable target, with no custom code. Multi-AZ, nightly batch copies and a single writer each fail the requirement for both Regions to take writes or the time target.
Want more on Multi-AZ and multi-Region high availability? The 5 full-length practice sets have 13 more Multi-AZ and multi-Region high availability questions, 375 in all, none of them in the free exam. One payment of $14.99, no subscription. See the practice sets →