Skip to content

AWS Certified Solutions Architect - Professional (SAP-C03) - Trade-off Scenarios

Original scenarios showing how constraints select the answer. They are not vendor exam questions.

Domain 1: Architecture

Many microservices, many accounts, minimal operations

Scenario: A company runs 40 services across 12 accounts on ECS and Lambda. Teams want service-to-service calls with authentication policies and request metrics, and no sidecars to manage. Analysis: the constraint is "no sidecars, many accounts and VPCs". VPC Lattice provides service networks with auth policies across VPCs and accounts. A self-run mesh adds operations; Service Connect is limited to ECS services in a namespace.

Hybrid DNS

Scenario: On-premises servers must resolve private hosted zone names, and VPC workloads must resolve corp.example names. Analysis: Route 53 Resolver inbound endpoint for on-premises to AWS, outbound endpoint with a forwarding rule for AWS to on-premises.

Domain 2: Security

Third-party vendor access

Scenario: A SaaS vendor must read cost data in your account. You must prevent another customer of the vendor from tricking the vendor into assuming your role. Analysis: a cross-account role with an external ID condition in the trust policy (confused deputy protection), with a least-privilege policy.

Region guardrail

Scenario: The compliance team requires that no resources be created outside two Regions, even by account administrators. Analysis: an SCP with a Deny on requests where aws:RequestedRegion is not in the approved list, excluding global services. An IAM policy inside each account could be removed by that account's admins.

Domain 3: Cost

Hidden transfer charge

Scenario: Private-subnet workloads move 30 TB a month to and from S3, and the NAT gateway data processing line is the largest networking cost. Analysis: add an S3 gateway VPC endpoint (no charge), routing S3 traffic around the NAT gateway.

Domain 4: Resilience

Matching DR to numbers

Scenario: An application needs an RTO of 15 minutes and an RPO of 1 minute, but the budget excludes running full capacity in a second Region. Analysis: warm standby: continuous replication (RPO) plus a scaled-down running environment that scales up on failover (RTO). Pilot light's longer startup may miss 15 minutes; active/active exceeds the budget.

Domain 5: Operations

Safe automated remediation

Scenario: An alarm triggers a Lambda function that restarts instances. During a regional dependency outage it restarts hundreds of instances repeatedly and worsens the problem. Analysis: add safety constraints: rate limits, a circuit breaker that stops remediation when failures persist, and limits on the blast radius per run.