Examkin

Soru

Zorluk: Çok zorMulti-AZ, Multi-Region Architectures and Disaster Recovery (DR)

An enterprise energy grid monitoring company runs its real-time telemetry processing application on AWS. The application consists of Amazon ECS Fargate tasks in a private VPC subnet in us-east-1 behind an Application Load Balancer (ALB). The database tier stores state in an Amazon Aurora PostgreSQL DB cluster. The data ingest rate is highly critical, and the company requires a cross-region disaster recovery (DR) solution in us-west-2 with a Recovery Time Objective (RTO) of 1515 minutes and a Recovery Point Objective (RPO) of 11 minute. The solution must minimize operational costs during normal operations. Which TWO actions should a solutions architect recommend to implement this DR strategy?

  1. Create an Amazon Aurora Global Database from the primary database cluster. In us-west-2, deploy a secondary DB cluster containing an Amazon Aurora Serverless v2 reader instance configured with a minimum capacity of 0.50.5 Aurora Capacity Units (ACUs).Cevap
  2. In us-west-2, provision the network infrastructure, the Application Load Balancer, and the Amazon ECS service with the desired task count set to 0. Create an automation workflow to scale the ECS service to the required production capacity during a failover.Cevap
  3. C
    In us-west-2, deploy a scaled-down Amazon ECS service running at 50% capacity, and configure an Amazon Route 53 latency-based routing policy to automatically distribute incoming traffic across both Regions during normal operations.
  4. D
    Configure Amazon Aurora to export hourly database snapshots to Amazon S3. Use S3 Cross-Region Replication (CRR) to replicate these snapshots to an S3 Glacier Flexible Retrieval vault in us-west-2, and write a script to restore the cluster during failover.
  5. E
    Create a cross-region Amazon RDS PostgreSQL read replica in us-west-2. Configure an AWS Lambda function to automatically promote this replica to a standalone primary database cluster using the RDS API during a primary Region outage.

Cevap

Configure an Amazon Aurora Global Database with an Aurora Serverless v2 reader in the secondary Region, and deploy the secondary ECS infrastructure with a task count of 0, scaling up only during failover.
Using an Amazon Aurora Global Database secondary cluster with a Serverless v2 reader instance allows replication to proceed with sub-second latency, ensuring the 1-minute RPO is met while keeping idle compute costs very low (0.50.5 ACUs). Setting the ECS service's desired task count to 0 in the secondary Region implements a Pilot Light pattern, which eliminates active ECS compute costs during normal operations. The infrastructure can be scaled up to target capacity within the 15-minute RTO during a failover event.

Adım Adım Çözüm

1
Determine the database configuration required to meet the 1-minute RPO across regions.
An Amazon Aurora Global Database secondary cluster in the target Region.
Aurora Global Database replication has typical latencies of under 1 second, meeting the 1-minute RPO requirement. To minimize idle cost, deploying an Aurora Serverless v2 reader with a minimum capacity of 0.50.5 ACUs provides replication support without paying for full-scale provisioned instances.
2
Determine the application compute configuration required to meet the 15-minute RTO while minimizing running costs.
An ECS service configured in the secondary Region with desired task count set to 0 (Pilot Light strategy).
Because ECS tasks can scale up from 0 to full capacity in under 5 minutes, keeping the ECS service task count at 0 during normal operations saves compute costs while easily fitting within the 15-minute recovery timeline.
3
Evaluate and rule out standard backup/restore or RDS replica promotion solutions.
Reject database snapshots replicated to S3 Glacier and standard RDS read replicas.
Snapshot restore times and Glacier retrieval delays exceed the 15-minute RTO, and snapshots fail the 1-minute RPO. Standard RDS read replicas are incompatible with Aurora clusters and lack the seamless managed global database failover mechanism.

Anahtar Kavram

Disaster recovery (DR) strategies, specifically Pilot Light database replication and scaled-down ECS Fargate deployments, to meet strict RTO/RPO targets while minimizing idle costs.
Bu soruyu puanla

Konular

Multi-AZ, Multi-Region Architectures and Disaster Recovery (DR)

Bu konudan daha fazla soru