Skip to content

AWS DevOps Engineer Pro (DOP-C02) - Exam Strategy

Cert-specific tactics. General study advice lives in study-strategies.md.

Format reminder

  • 75 scored questions, 180 minutes
  • Pass mark ~750 / 1000 (~75%)
  • Multiple choice + multiple response
  • Heavy on practical scenarios; lighter on architecture trade-offs than SAP-C02

Time management math

180 / 75 = 2.4 min/question. DOP-C02 questions tend to be shorter than SAP-C02 - many can be answered in 60-90 seconds if you recognize the pattern. Bank time on the easy ones for the longer pipeline scenarios.

Pace: Q25 by 60 min, Q50 by 120 min, finish by 165 min, leave 15 min for flagged review.

Top traps for THIS exam

  1. CodeDeploy deployment styles: blue/green vs in-place; Lambda canary/linear/all-at-once presets (Canary10Percent10Minutes, Linear10Percent5Minutes, etc.). Memorize the named presets - they appear verbatim in answers.

  2. Pipeline cross-account: the right answer almost always involves sts:AssumeRole with cross-account roles, never long-lived credentials. CodePipeline action's RoleArn field is the mechanism.

  3. SSM agent + private subnet: three VPC endpoints (ssm, ec2messages, ssmmessages) allow SSM agent to operate without internet egress. This appears in both Patch Manager and Run Command questions.

  4. EventBridge vs CloudWatch Events - the same service, different brand. Don't second-guess; pick whichever the question phrases.

  5. Auto-rollback alarms: CodeDeploy supports up to 10 rollback alarms per deployment group. The alarm pattern (CloudWatch alarm → CodeDeploy rollback) is heavily tested.

  6. CloudFormation StackSets: the AWS-managed permission model (service-managed permissions via Organizations) is the modern answer; self-managed permissions are legacy and require setting up trust roles per account manually.

  7. Systems Manager Incident Manager is the AWS-recommended IR product (introduced 2021); shows up as the right answer to many "automated incident response" questions.

  8. Container deploy patterns:

  9. ECS rolling update = simple deploys, no canary
  10. CodeDeploy blue/green for ECS = canary + auto-rollback, requires two target groups
  11. ECS service revision shift via task set = also blue/green, more advanced

  12. Know which pattern goes with which constraint

  13. CodeBuild local testing with aws codebuild start-build and the use of buildspec.yml versions (0.2 has phases, artifacts, cache, env).

  14. AWS Backup vs Lifecycle Manager (DLM): Backup is multi-service (RDS, EBS, EFS, DynamoDB, etc.) with central audit; DLM is EBS snapshots only. Pick Backup for cross-service.

Common high-yield topics easy to miss

  • AWS Proton (managed self-service for platform engineering)
  • AWS App2Container (containerizing Java + .NET workloads)
  • AWS Copilot (CLI for ECS/App Runner deployments)
  • Lambda destinations (vs DLQs - destinations are async-only, both success and failure)
  • Step Functions express workflows vs standard (express = high volume, low duration, no exec history)
  • AWS Config conformance packs (multi-rule deployment via CFN)
  • Aurora Backtrack (point-in-time within minutes without restore)
  • S3 Replication Time Control (RTC) for SLA-backed cross-region replication
  • AWS Resilience Hub (assess and improve resilience score)

When you're stuck on a question

  1. Eliminate manual / non-managed answers - "an engineer SSHs in and runs..." is rarely correct.
  2. Eliminate "build with Lambda" if a managed service does it - Lambda glue is fine when nothing native exists, but most DOP-C02 problems have a managed service answer.
  3. Map to the deciding constraint - is it speed of rollback? minimize downtime? cost? operational overhead?
  4. Pattern-match the named features - blue/green, canary, in-place, rolling - these are tested as named patterns.
  5. Flag and move on - 5 minutes max per question.

Day-of logistics

  • 180 min, 75 questions: similar pacing to SAP-C02
  • Bring two government IDs to the testing center
  • Online proctored: clear room, single monitor, no books, no second screens
  • The provided whiteboard / scratch is minimal - don't rely on it for math, do mental math

After the exam

Pass: Professional cert is valid 3 years; recert by passing again or any other AWS Pro/Specialty.

Fail: Score breakdown by domain. If you failed close to the cut, regroup in 4-6 weeks; you can retake after 14 days. Most failures involve missing the SDLC Automation domain (CodePipeline / CodeDeploy / CodeBuild trivia) or the Incident Response domain (Incident Manager + EventBridge patterns).

Patterns to internalize

  • "Zero-downtime deploy" + "auto rollback" = CodeDeploy blue/green with alarm
  • "Cross-account pipeline" = AssumeRole pattern
  • "Patch at scale" = SSM Patch Manager
  • "Centralized observability across accounts" = CloudWatch cross-account observability
  • "Automated remediation" = EventBridge → Step Functions or Incident Manager runbook
  • "Drift detection" = CloudFormation native drift APIs
  • "Self-service infrastructure" = Service Catalog or Proton
  • "Canary Lambda deploy" = SAM DeploymentPreference: Canary10Percent10Minutes