Examkin

Question

Difficulty: HardHigh Availability and Disaster Recovery Design

A digital banking platform is designing a new multi-region credit card processing system on AWS. The system must be deployed in us-east-1 (Primary) and us-west-2 (Secondary). The architecture requires a Recovery Point Objective (RPO) of less than 5 seconds and a Recovery Time Objective (RTO) of less than 15 minutes. During normal operations, global users must be routed to the region with the lowest latency, and local reads must be served from the nearest region to minimize response times. If the application in the primary region fails, traffic must failover to the secondary region automatically within 5 minutes. Outbound payment network calls from the application instances must remain highly available even if a single Availability Zone experiences an outage. The design should minimize operational complexity while meeting these requirements.

Which of the following architectures meets these requirements?

  1. A
    Deploy the application on Amazon ECS in both Regions behind Application Load Balancers (ALBs). Configure Amazon Route 53 with Latency-based routing records for both ALBs, with Route 53 health checks associated with each record. Deploy an Amazon Aurora Global Database with the writer cluster in us-east-1 and a reader cluster in us-west-2. Deploy a single NAT Gateway in each VPC in the first Availability Zone, and update all private subnet route tables to route outbound internet traffic through this single NAT Gateway.
  2. B
    Deploy the application on Amazon ECS in us-east-1 behind an Application Load Balancer (ALB). Configure Amazon Route 53 with Failover routing to direct traffic to us-east-1 as primary and an Amazon S3 static website hosting a maintenance page in us-west-2 as secondary. Use Amazon Aurora Serverless v2 in us-east-1, take hourly database snapshots, copy them to us-west-2, and use AWS CloudFormation to restore the environment during a disaster.
  3. Deploy the application on Amazon ECS in both Regions behind Application Load Balancers (ALBs). Configure Amazon Route 53 with Latency-based routing records for both ALBs, with Route 53 health checks associated with each record. Deploy an Amazon Aurora Global Database with the writer cluster in us-east-1 and a reader cluster in us-west-2. Deploy NAT Gateways in every Availability Zone within each Region's VPC, updating the private subnet route tables to route outbound internet traffic through the local NAT Gateway in the same zone.Answer
  4. D
    Deploy the application on Amazon ECS in both Regions behind Application Load Balancers (ALBs). Configure Amazon Route 53 with Latency-based routing records for both ALBs, with Route 53 health checks associated with each record. Deploy Amazon RDS for PostgreSQL in us-east-1, and configure an Amazon RDS Multi-AZ standby instance in us-west-2, instructing the ECS tasks in the secondary region to query the standby instance directly to serve local read requests. Deploy NAT Gateways in every Availability Zone within each Region's VPC.

Answer

Deploy the application on Amazon ECS in both Regions behind ALBs, configure Route 53 Latency-based routing with health checks, use Amazon Aurora Global Database with a primary in us-east-1 and a secondary in us-west-2, and deploy NAT Gateways in every Availability Zone in each Region's VPC.
The correct architecture uses Amazon Aurora Global Database to achieve cross-region replication lag of less than a second, while allowing the reader cluster in the secondary region to serve local reads. Amazon Route 53 with Latency-based routing and health checks ensures users are sent to the nearest healthy region and handles automatic failover. High availability of outbound traffic is achieved by deploying redundant NAT Gateways across all Availability Zones.

Step-by-Step Solution

1
Select a multi-region database solution that supports sub-second cross-region replication and local read scaling to meet the 5-second RPO and local read latency requirements.
Amazon Aurora Global Database provides asynchronous cross-region replication (typically under 1 second) and allows the secondary cluster to serve local reads, fulfilling both database requirements.
Standard multi-region snapshot copies have too high an RPO, and RDS Multi-AZ standbys are located in the same region and do not serve read traffic.
2
Configure the DNS routing policy to route users based on latency and automatically failover when a region is unhealthy.
Amazon Route 53 Latency-based routing records associated with health checks automatically direct users to the region with the lowest latency and stop routing traffic to a degraded region within minutes of a health check failure.
Without health checks, Route 53 cannot perform automatic failover, and static failover policies do not direct active users to the lowest-latency healthy region.
3
Ensure outbound connectivity from private subnets is highly available in the event of an Availability Zone outage.
Deploying a dedicated NAT Gateway in each Availability Zone prevents a single point of failure. If an Availability Zone fails, only that zone's resources are affected, and resources in other zones continue using their local NAT Gateways.
Routing all private subnet traffic through a single NAT Gateway in a single Availability Zone creates a single point of failure for the entire VPC.

Key Concept

High Availability and Disaster Recovery Design for Multi-Region Architectures
Rate this question

Topics

High Availability and Disaster Recovery Design

More questions from this topic