AWS Certified DevOps Engineer – Professional (DOP-C02)
The professional DevOps exam. It covers CI/CD, infrastructure as code, resilience, monitoring, incident response and compliance automation.
Verified against official sources on 2026-11-17.
AWS Certified DevOps Engineer – Professional (DOP-C02) checks that you can provision, operate and manage distributed systems on AWS. That means continuous delivery, automated security and compliance, monitoring and logging, self-healing systems, and automated operations. Passing it also renews DVA and SOA.
At a glance
| Code | DOP-C02 |
| Level | Professional |
| Intended audience | 2 or more years provisioning, operating and managing AWS environments, plus SDLC and programming or scripting experience |
| Questions | 75 (65 scored + 10 unscored) |
| Time | 180 minutes |
| Passing score | 750 (scaled 100–1000) |
| Price | USD 300 |
| Question types | Multiple choice, multiple response |
| Valid for | 3 years. Renew by passing the latest DOP exam, or with the paid Skill Builder maintain option |
| Out of scope | Advanced networking, deep security recommendations, database design and tuning, full-stack development |
Official domains and weightings
| Domain | Weight |
|---|---|
| 1. SDLC Automation | 22% |
| 2. Configuration Management and IaC | 17% |
| 3. Resilient Cloud Solutions | 15% |
| 4. Monitoring and Logging | 15% |
| 5. Incident and Event Response | 14% |
| 6. Security and Compliance | 17% |
Key AWS services tested
- CI/CD: CodePipeline, CodeBuild, CodeDeploy (EC2, ECS and Lambda deployment types), CodeArtifact, ECR, third-party CI (for example GitHub Actions) using IAM OIDC federation
- IaC and configuration: CloudFormation (change sets, StackSets, custom resources, helper scripts), CDK, SAM, Systems Manager (State Manager, Parameter Store, Automation), AWS AppConfig, Service Catalog, Control Tower and Organizations
- Resilience: Auto Scaling (lifecycle hooks, warm pools), ELB, Route 53, Aurora and DynamoDB global tables, AWS Backup, Fault Injection Service
- Monitoring: CloudWatch (Logs, metric filters, Logs Insights, Synthetics, anomaly detection), X-Ray, the OpenTelemetry distro, EventBridge, AWS Health
- Incident response: EventBridge rules, SNS, Lambda, Systems Manager OpsCenter and Incident Manager
- Security and compliance: IAM, Config (conformance packs, remediation), Security Hub, GuardDuty, Inspector, KMS, Secrets Manager, SCPs
What you already know vs what you must learn
| You already know | You must learn |
|---|---|
| Azure DevOps or GitHub Actions YAML pipelines, branching, PR policies | CodePipeline, CodeBuild and CodeDeploy behaviour in detail (appspec hooks, deployment configurations, rollbacks) |
| Bicep or ARM, maybe Terraform | CloudFormation at scale: StackSets, change sets, drift detection, custom resources. CDK in C# |
| Release approvals and audit evidence | Multi-account deployment: cross-account roles, artifact promotion, KMS keys for artifacts |
| Health checks in ASP.NET Core | Self-healing: Auto Scaling lifecycle hooks, EventBridge-driven remediation, ECS deployment circuit breakers |
| Application Insights alerts | Automating responses to CloudWatch alarms, Config findings and GuardDuty events |
Preparation order and study plan (about 60–80 hours over 6–8 weeks, after SOA-C03)
| Week | Focus | Hours |
|---|---|---|
| 1 | Exam guide. Module 09 pipelines (OIDC, CodePipeline, CodeBuild) | 10 |
| 2 | CodeDeploy for EC2, ECS (blue/green) and Lambda. Rollback configuration | 10 |
| 3 | Module 10: CloudFormation advanced, StackSets, CDK in C#, cdk-nag | 10 |
| 4 | Domain 3: Module 12 resilience. Fault Injection Service | 8 |
| 5 | Domains 4 and 5: Module 11 monitoring, EventBridge automation, incident runbooks | 8–10 |
| 6 | Domain 6: Config conformance packs, Security Hub, SCPs, secrets in pipelines | 8 |
| 7–8 | Free Exam Prep course, the 20-question set, this page's questions, long scenario practice, mock exams | 10–12 |
Mapped platform modules
- Module 09 — DevOps & CI/CD: Domain 1
- Module 10 — Infrastructure as Code: Domain 2
- Module 12 — HA & DR: Domain 3
- Module 11 — Monitoring: Domains 4 and 5
- Module 05 — Security and Module 15 — Governance: Domain 6
- Module 17 — Playbooks: incident runbooks
Practical skills to build
- GitHub Actions deploying to AWS through an OIDC role whose trust policy is restricted to one repository and branch
- A CodePipeline with build, unit tests, a security scan, a staging deploy, manual approval, and a production deploy to ECS with blue/green and automatic rollback
- A CDK (C#) app with cdk-nag checks that fail the build on security findings
- StackSets that deploy a baseline (Config rules, IAM roles) to every account in an OU, including new ones
- An EventBridge rule that turns a GuardDuty or Config finding into automatic remediation, plus an SNS notification
Exam-prep checklist
- I know the CodeDeploy deployment configurations and the lifecycle hook order for EC2, ECS and Lambda
- I can set up cross-account pipelines (artifact bucket KMS key, cross-account roles)
- I know CloudFormation StackSets (self-managed and service-managed), change sets,
DeletionPolicy,UpdateReplacePolicy, and custom resources - I can use Auto Scaling lifecycle hooks and warm pools
- I can build an event-driven remediation flow (EventBridge to Lambda or SSM Automation)
- I know how to aggregate logs and metrics across accounts
- I can keep secrets out of pipelines and use OIDC instead of long-lived keys
- I can work through a long multi-paragraph scenario within about 2.4 minutes per question
Official links
Practice questions
These 5 questions are original, banking-flavoured scenarios. They aren't real exam questions.
The bank's security policy bans long-lived AWS access keys in CI systems. A team uses GitHub Actions to deploy a .NET API to a production account, and only the main branch of one repository may deploy. What should the DevOps engineer set up?
- A. An IAM user with an access key stored as a GitHub encrypted secret, rotated every 90 days
- B. An IAM OIDC identity provider for GitHub, and a deploy role whose trust policy checks
audissts.amazonaws.comand restrictssubtorepo:<org>/<repo>:ref:refs/heads/main(or to a protected GitHub environment) - C. An OIDC provider and a role whose trust policy checks only the
audclaim - D. The root user's access keys stored in an encrypted secret
Correct answer: B
Explanation: With OIDC, each workflow run gets a short-lived token that it exchanges for temporary AWS credentials (AssumeRoleWithWebIdentity), so there are no stored keys. Security depends on the trust policy conditions. Restricting sub to the repository and branch, or to an environment that has required reviewers, stops any other repository from assuming the role. Keep the role's permissions to the minimum needed to deploy.
Why the others are wrong:
- A: That's still a long-lived key, which the policy bans, and rotating it doesn't remove the risk of it leaking.
- C: Without a
subcondition, any GitHub repository could assume the role. This misconfiguration is well known and dangerous. - D: Root credentials must never be used for automation.
A customer-facing ASP.NET Core API runs on ECS Fargate behind an ALB. The bank needs releases that (1) let testers check the new version on a separate test listener before production traffic moves, (2) move traffic all at once or in steps, and (3) roll back automatically if 5xx errors spike. Which approach fits?
- A. An ECS rolling update with minimum healthy percent set to 100
- B. Blue/green deployment with two target groups, a production listener and a test listener, and CloudWatch alarms configured to trigger automatic rollback
- C. Delete the service and recreate it with the new task definition
- D. Update the task definition and restart tasks manually at night
Correct answer: B
Explanation: Blue/green runs the new (green) task set next to the old (blue) one. The test listener sends tester traffic to green before the switch. Traffic then moves according to the deployment configuration (all at once, canary or linear), and an alarm triggers automatic rollback to blue. The DOP-C02 guide covers this through CodeDeploy blue/green for ECS. Amazon ECS has since added built-in blue/green deployments as well. ⚠ VERIFY which option current exam questions expect.
Why the others are wrong:
- A: Rolling updates replace tasks in place. There's no separate test listener, and rollback depends on the deployment circuit breaker rather than a managed traffic switch.
- C: Recreating the service causes an outage.
- D: Manual restarts have no automated checks or rollback and don't meet change-control expectations.
The cloud platform team must deploy a standard set of AWS Config rules and a read-only audit IAM role to every account in the Workloads OU, including accounts created in future, without running deployments by hand. What should they use?
- A. CloudFormation nested stacks deployed into each account by hand
- B. CloudFormation StackSets with service-managed permissions (Organizations integration), targeting the OU with automatic deployment turned on
- C. A CDK app that a developer runs against each account from a laptop
- D. AWS Config aggregators
Correct answer: B
Explanation: Service-managed StackSets use Organizations to create the required roles for you. With automatic deployment, any account that joins the target OU gets the stack, and an account that leaves it has the stack removed (if you configure that). This is how you keep a baseline consistent at scale. Control Tower customisations build on the same mechanism.
Why the others are wrong:
- A: Nested stacks organise templates within a single stack deployment. They don't spread across accounts.
- C: Manual runs from a laptop can't be audited, depend on one person, and miss new accounts.
- D: Aggregators collect compliance data across accounts. They don't deploy rules or roles.
EC2 instances in an Auto Scaling group sometimes fail health checks and get replaced. By the time engineers investigate, the instances and their local application logs are gone. How can the team capture diagnostics automatically before termination?
- A. Turn off health checks on the Auto Scaling group
- B. Add an Auto Scaling lifecycle hook for
EC2_INSTANCE_TERMINATING. An EventBridge rule on the hook event runs an SSM Automation or Run Command document that copies the logs to S3, then completes the lifecycle action - C. Increase the health check grace period to 1 hour
- D. Take a daily AMI of every instance
Correct answer: B
Explanation: A terminating lifecycle hook pauses the instance in the Terminating:Wait state until the action is completed or the timeout is reached. EventBridge receives the event and triggers automation that collects diagnostics. In the longer term, stream logs continuously to CloudWatch Logs with the agent, so nothing is lost even if an instance fails suddenly.
Why the others are wrong:
- A: Turning off health checks leaves unhealthy instances serving traffic.
- C: A longer grace period delays replacement of failed instances and doesn't capture any logs.
- D: Daily AMIs miss the moment of failure and cost money.
The bank's change management policy requires that (1) production deployments are approved by a named release manager, and (2) auditors can later see who approved what, and when. The team uses CodePipeline. Which two measures meet these requirements?
- A. Add a manual approval action before the production stage, send notifications through SNS, and allow
codepipeline:PutApprovalResultonly for the release-manager role - B. Rely on CloudTrail, which records the
PutApprovalResultAPI call with the approver's identity, and keep the logs in the central log-archive account - C. Treat merging a pull request to
mainas production approval - D. Ask release managers to approve by replying to an email
- E. Turn off the production stage between releases
Correct answers: A and B
Explanation: The manual approval action stops the pipeline until an authorised person approves or rejects it, and IAM limits who that can be. CloudTrail records the API call that made the approval, with the caller's identity and a timestamp. Stored immutably, that's the audit evidence. Add comments and links to the change ticket in the approval to make the record more complete.
Why the others are wrong:
- C: Code review isn't the same as production change approval. Banks usually require them to be separate (segregation of duties).
- D: An email reply isn't tied to the pipeline and is hard to audit.
- E: Turning stages off is an operational hack. It doesn't record any approval.
Independent learning resource — not affiliated with, endorsed or sponsored by Amazon Web Services. AWS and related marks are trademarks of Amazon.com, Inc. or its affiliates. All banking data in examples is fictional. Sample architectures are for learning and are not automatically compliant with GDPR, PCI DSS or any banking regulation.
AWS Certified CloudOps Engineer – Associate (SOA-C03)
The operations exam, formerly SysOps Administrator. It covers monitoring, reliability, automation, security, networking and troubleshooting on AWS.
AWS Certified Solutions Architect – Professional (SAP-C03)
The hardest exam in the roadmap. It covers complex, multi-account, hybrid, secure, resilient and AI-integrated architectures.