Highest-weight domain. Covers the broadest range of AWS services. Expect ~19 questions focused on architecture decisions, trade-offs, and integration patterns.

What This Domain Tests

• Selecting the right compute, storage, database, and networking services for given requirements
• Designing scalable, highly available, and fault-tolerant architectures
• Applying security, encryption, and access controls to new workloads
• Choosing between serverless, container, and EC2-based architectures
• Designing event-driven and decoupled architectures
• Applying IaC best practices with CloudFormation and CDK

Subdomain Breakdown

Task names aligned to the official SAP-C02 exam guide. The guide may change; recheck before exam day.

Files in This Domain

Compute/

• EC2-Advanced-General.md — placement groups, Spot, Reserved, instance types
• ECS-General.md — Fargate, task definitions, service discovery
• EKS-General.md — managed node groups, Fargate profiles, networking
• Lambda-General.md — concurrency, layers, destinations, SnapStart
• AutoScaling-General.md — scaling policies, predictive, warm pools

Networking/

• CloudFront-General.md — distributions, origins, cache behaviors, security
• APIGateway-General.md — REST vs HTTP vs WebSocket, throttling, caching
• ELB-General.md — ALB, NLB, GWLB — when to use each
• GlobalAccelerator-General.md — anycast, traffic dials, vs CloudFront

Storage/

• S3-Advanced-General.md — storage classes, lifecycle, replication, encryption
• EFS-General.md — shared file storage, performance modes, replication
• FSx-General.md — FSx for Windows, Lustre, NetApp ONTAP, OpenZFS
• EBS-General.md — volume types, snapshots, encryption, multi-attach

Databases/

• RDS-Aurora-General.md — Multi-AZ, read replicas, Aurora Global, Serverless
• DynamoDB-General.md — partitions, GSI, streams, DAX, Global Tables
• ElastiCache-General.md — Redis vs Memcached, caching patterns
• OpenSearch-General.md — search, log analytics, domain design
• Purpose-Built-Database-Selector.md — DocumentDB, Keyspaces, Neptune, Redshift, Timestream and selection cues

Serverless/

• StepFunctions-General.md — Standard vs Express, state types, error handling
• EventBridge-General.md — event buses, rules, pipes, scheduler
• SQS-SNS-General.md — queues, topics, fan-out, DLQ, FIFO
• Kinesis-General.md — Data Streams, Firehose, Data Analytics

Standalone

• MSK-General.md — managed Kafka, use cases vs Kinesis
• CloudFormation-General.md — stacks, StackSets, drift, nested stacks
• CDK-General.md — constructs, stacks, pipelines, vs CloudFormation
• Deployment-Strategies-General.md — CodePipeline, CodeBuild, CodeDeploy, rollout and rollback patterns

Key Architecture Patterns

Three-Tier Web Application

Internet → CloudFront → ALB → ECS/EC2 (Auto Scaling) → Aurora Multi-AZ
                                    ↓
                              ElastiCache (Redis)

Serverless API

Client → API Gateway → Lambda → DynamoDB
                    ↓
              SQS → Lambda → downstream processing

Event-Driven Architecture

Producer → EventBridge → Rules → Multiple consumers
                              ├── Lambda (real-time processing)
                              ├── SQS (buffered processing)
                              └── Step Functions (orchestration)

Multi-Region Active-Active

Route 53 / Global Accelerator
      ├── Region A: ALB → ECS → Aurora Global primary (writer)
      └── Region B: ALB → ECS → Aurora Global secondary (read-only)
                              └── optional write forwarding to primary

For multi-Region writes, use a data store designed for that model (for example,
DynamoDB Global Tables) and design for its consistency/conflict semantics.

Exam Tips

• Decouple with SQS/SNS/EventBridge when asynchronous processing, burst buffering, or independent consumers meet the requirements; account for retries and eventual consistency
• Serverless can fit variable or event-driven workloads; validate runtime, duration, concurrency, data, latency, and total-cost constraints
• Aurora vs RDS is workload-specific: compare engine compatibility, HA/replication needs, features, performance, operations, and cost
• DynamoDB fits access-pattern-driven key-value/document workloads; design partition keys, indexes, throughput, consistency, and hot-key behavior
• CloudFront for static content and global distribution — not just CDN, also security (WAF, Shield)
• ECS vs EKS — ECS (use the Fargate launch type for serverless compute) for AWS-native simplicity, EKS for Kubernetes compatibility
• Auto Scaling — use health checks and multiple AZs where availability requirements call for them
• Cost optimization: Spot for fault-tolerant, Savings Plans for steady-state, serverless for variable