2026.06.03 2026.06.03

AWS DR Design for SAA-C03: RTO/RPO and the 4 Recovery Strategies

swiftwand

Domain 3 of the AWS Certified Solutions Architect – Associate (SAA-C03), Design High-Performing Architectures, carries 24 percent of the scored content. Performance design tests one skill: reverse-mapping from a workload’s characteristics to the optimal service. Across four areas – compute, storage, database, and networking – you must choose the service that best fits the requirement.

Compute performance design – EC2, Lambda, and containers
Storage performance design – EBS, S3, and EFS
Database performance design – RDS, Aurora, DynamoDB, and ElastiCache
Network performance design – CloudFront and Global Accelerator
Summary – the decision axes of high-performing design

忍者AdMax

Compute performance design – EC2, Lambda, and containers

At the compute layer, the choice follows the workload. Always-on with predictable load points to EC2; event-driven, short-duration work points to Lambda; containerized microservices point to ECS/EKS. A frequent decision point is that Lambda has a 15-minute execution limit, making it unsuitable for long-running batch jobs.

EC2 instance type selection matters too. Match general purpose (M family), compute optimized (C family), memory optimized (R family), and storage optimized (I family) to the workload. On the exam, CPU-heavy scientific computing points to the C family; an in-memory database points to the R family.

Storage performance design – EBS, S3, and EFS

Reverse-map the storage layer from the access pattern. Fast block storage for a single instance means EBS; shared access from multiple instances means EFS; object storage with high durability means S3. Choose EBS volume types such as gp3 and io2 from your IOPS requirement.

For large data transfers, distinguish AWS DataSync from Snowball. Online transfer points to DataSync; petabyte-scale offline transfer points to Snowball.

Database performance design – RDS, Aurora, DynamoDB, and ElastiCache

The database layer is the most frequently tested on SAA-C03. Relational points to RDS or Aurora; ultra-low-latency NoSQL points to DynamoDB; a caching layer points to ElastiCache. When read load is high, the exam looks for read replicas to distribute it.

DynamoDB offers single-digit millisecond latency and automatic scaling. Pair it with DynamoDB Accelerator (DAX) for microsecond-level caching. On the exam, automatically handling unpredictable spikes or serverless NoSQL usually points to DynamoDB.

Network performance design – CloudFront and Global Accelerator

At the network layer, the exam tests service choices for reducing latency. Delivering static and dynamic content points to CloudFront, a CDN; globally accelerating TCP/UDP points to Global Accelerator. CloudFront caches content at edge locations and serves it from a place close to the user.

On the exam, delivering static content to users worldwide at low latency points to CloudFront; accelerating a non-HTTP global application points to Global Accelerator.

Summary – the decision axes of high-performing design

SAA-C03 high-performing design comes down to reverse-mapping from a workload’s characteristics to the optimal service. Choose compute from the load pattern, storage from the access pattern, database from the data model and latency requirement, and networking from the delivery footprint. Organize each service by requirement-to-service, and Domain 3 becomes a source of points.

Performance and cost trade off against each other. Studying this alongside the Domain 4 cost-optimization article builds the judgment to hold performance while keeping cost down.

ブラウザだけでできる本格的なAI画像生成【ConoHa AI Canvas】