AWS Certified CloudOps Engineer – Associate (SOA-C03)
The operations exam, formerly SysOps Administrator. It covers monitoring, reliability, automation, security, networking and troubleshooting on AWS.
Verified against official sources on 2026-11-17.
AWS Certified CloudOps Engineer – Associate (SOA-C03) checks that you can deploy, manage and operate workloads on AWS. That means monitoring and fixing problems, keeping systems available and recoverable, automating provisioning, applying security controls, and troubleshooting networks. It used to be called SysOps Administrator – Associate.
The name change
SOA-C03 replaced SOA-C02 on 2025-09-30, and the certification is now called CloudOps Engineer – Associate. If you already hold SysOps Administrator, it stays valid for its full 3 years, and you don't have to take the new exam. Only people who pass SOA-C03 get the new title.
At a glance
| Code | SOA-C03 |
| Level | Associate |
| Intended audience | 1 year of deploying, managing, troubleshooting, networking and securing workloads on AWS, plus 1 year in an operations role |
| Questions | 65 (50 scored + 15 unscored) |
| Time | 130 minutes |
| Passing score | 720 (scaled 100–1000) |
| Price | USD 150 |
| Question types | Multiple choice, multiple response |
| Valid for | 3 years. Renewed by passing DOP-C02, or with the paid Skill Builder maintain option |
| Out of scope | Designing distributed architectures, CI/CD pipelines, or hybrid and multi-VPC networks. Developing software. Defining governance requirements. Billing and TCO analysis |
Official domains and weightings
| Domain | Weight |
|---|---|
| 1. Monitoring, Logging, Analysis, Remediation, and Performance Optimization | 22% |
| 2. Reliability and Business Continuity | 22% |
| 3. Deployment, Provisioning, and Automation | 22% |
| 4. Security and Compliance | 16% |
| 5. Networking and Content Delivery | 18% |
Compared with SOA-C02: the old "Cost and Performance Optimization" domain was merged into Domain 1. New topics include configuring the CloudWatch agent for EC2, ECS and EKS, managing stacks with CloudFormation and the CDK, enforcing compliance rules such as allowed Regions and services, and CloudWatch network monitoring. S3 static website hosting was removed.
Key AWS services tested
- Monitoring: CloudWatch (metrics, the agent, Logs and Logs Insights, alarms, composite alarms, dashboards, Synthetics, network monitoring), EventBridge, CloudTrail, X-Ray, AWS Health
- Operations automation: Systems Manager (Session Manager, Patch Manager, Run Command, Automation, Parameter Store, State Manager, Inventory), Config with remediation
- Reliability: EC2 Auto Scaling, Elastic Load Balancing, RDS Multi-AZ and read replicas, Aurora, DynamoDB, ElastiCache, AWS Backup, Route 53 health checks and failover
- Provisioning: CloudFormation (StackSets, drift detection), CDK, Image Builder, AMIs, ECS and EKS basics
- Security: IAM, Organizations and SCPs, KMS, Secrets Manager, ACM, Security Hub, GuardDuty, Inspector, Trusted Advisor
- Networking: VPC (endpoints, peering, Transit Gateway basics, Flow Logs, Reachability Analyzer), Site-to-Site VPN, Route 53 (including Resolver), CloudFront, WAF, Network Firewall
What you already know vs what you must learn
| You already know | You must learn |
|---|---|
| Reading logs, investigating production incidents, Windows Server administration | The CloudWatch agent, metric filters, alarms tied to automated actions |
| Patching windows and change control in banks | Systems Manager: Patch Manager, Maintenance Windows, Session Manager (no RDP or SSH ports) |
| SQL Server backups and restore drills | AWS Backup plans, cross-account and cross-Region copies, Vault Lock |
| ARM, Bicep or PowerShell DSC | CloudFormation: change sets, drift detection, StackSets, stack policies |
| "It works on my machine" debugging | VPC troubleshooting: route tables, security groups, NACLs, endpoints, Flow Logs, Reachability Analyzer |
Preparation order and study plan (about 35–45 hours over 4–5 weeks, after SAA and DVA)
| Week | Focus | Hours |
|---|---|---|
| 1 | Exam guide and the C02→C03 comparison. Module 11 (CloudWatch agent, alarms, Logs Insights) | 8–10 |
| 2 | Module 12 (Auto Scaling, Backup, Route 53 failover). Systems Manager labs | 8–10 |
| 3 | Module 10 (CloudFormation, CDK, StackSets, drift). Module 15 Config and SCP compliance rules | 8 |
| 4 | Module 02 troubleshooting labs (endpoints, Flow Logs, Reachability Analyzer) | 6–8 |
| 5 | Free Exam Prep course, the 20-question set, this page's questions, a mock exam | 6 |
Mapped platform modules
- Module 11 — Monitoring, logging & troubleshooting: Domain 1
- Module 12 — HA & DR: Domain 2
- Module 10 — Infrastructure as Code: Domain 3
- Module 09 — DevOps & CI/CD: deployment strategies
- Module 05 — Security and Module 15 — Governance: Domain 4
- Module 02 — Networking: Domain 5
- Module 17 — Playbooks: ECS restarts, 502/503/504 errors, unexpected bills
Practical skills to build
- Install the CloudWatch agent with Systems Manager and publish memory and disk metrics from a Windows instance
- Patch a fleet with Patch Manager in a Maintenance Window and report compliance
- Create an AWS Backup plan with a cross-account copy, then restore from it and time the restore
- Detect CloudFormation drift, then fix it
- Diagnose "ECS task can't pull its image" in a private subnet using Flow Logs and Reachability Analyzer
- Auto-remediate a non-compliant S3 bucket with a Config rule and an SSM Automation runbook
Exam-prep checklist
- I know which EC2 metrics need the CloudWatch agent (memory, disk space) and how to install it at scale
- I can build alarms that trigger Auto Scaling, SNS or Systems Manager actions, and composite alarms
- I know Patch Manager, Maintenance Windows, Session Manager and Automation runbooks
- I can design backup and restore with AWS Backup and test it
- I can troubleshoot connectivity: route tables, SG versus NACL, endpoints, DNS resolution
- I know CloudFormation change sets, drift detection,
DeletionPolicyand StackSets - I can configure Route 53 health checks and failover routing
- I can enforce Region and service restrictions with SCPs and Config
Official links
- Exam guide (SOA-C03)
- Comparison of SOA-C02 and SOA-C03
- Exam product page
- Skill Builder Exam Prep plan (SOA-C03)
Practice questions
These 5 questions are original, banking-flavoured scenarios. They aren't real exam questions.
A legacy .NET Framework service runs on EC2 Windows instances. During a month-end batch the service slowed down, and the team suspects memory pressure. In CloudWatch they can see CPU and network metrics, but no memory or disk-space metrics. What should they do?
- A. Turn on EC2 detailed monitoring
- B. Install and configure the CloudWatch agent, for example through Systems Manager, to collect memory and logical-disk performance counters
- C. Turn on VPC Flow Logs
- D. Create a CloudTrail trail for EC2 events
Correct answer: B
Explanation: EC2's built-in metrics come from the hypervisor, which can't see inside the guest operating system, so memory and disk space aren't included. The CloudWatch agent collects OS-level metrics (Windows performance counters such as Memory % Committed Bytes In Use and LogicalDisk % Free Space) and logs. Store the agent configuration in Parameter Store and roll it out with Systems Manager so that every instance reports the same way.
Why the others are wrong:
- A: Detailed monitoring increases how often the existing hypervisor metrics are reported (every minute instead of every 5 minutes). It doesn't add memory.
- C: Flow Logs record network traffic metadata, not memory.
- D: CloudTrail records API calls, not operating system performance.
A bank's operations team has to patch 120 Windows EC2 instances every month in an approved change window, and give auditors a compliance report showing which instances are missing patches. The team wants as little manual effort as possible. What should they use?
- A. Log in to each instance over RDP and run Windows Update
- B. Systems Manager Patch Manager with a patch baseline, patch groups based on tags, and a Maintenance Window. Review compliance in Systems Manager (or aggregate it in Security Hub)
- C. Amazon Inspector
- D. Rebuild every instance from the latest marketplace AMI each month
Correct answer: B
Explanation: Patch Manager applies a baseline (which patches are approved, with optional automatic approval delays) to instances selected by tag. Maintenance Windows make patching run only in the approved window. The patch compliance data gives the audit evidence. Session Manager means no inbound RDP port has to be open.
Why the others are wrong:
- A: Manual work doesn't scale, can't be audited easily, and needs RDP access.
- C: Inspector finds vulnerabilities, including missing patches, but doesn't install patches.
- D: Rebuilding from images is a valid immutable approach, but marketplace AMIs won't include your application or configuration. Use a pipeline such as EC2 Image Builder if you take that route.
After a ransomware exercise, the bank requires that RDS and EBS backups be (1) copied to a separate AWS account and a second Region, and (2) impossible to delete early, even by administrators, for 35 days. What meets both requirements with the least custom work?
- A. A Lambda function that copies snapshots on a schedule
- B. An AWS Backup plan with cross-account and cross-Region copy rules to a vault in a dedicated backup account, with AWS Backup Vault Lock in compliance mode and a minimum retention of 35 days
- C. RDS automated backups with a 35-day retention period
- D. S3 Cross-Region Replication of database exports
Correct answer: B
Explanation: AWS Backup centralises backup schedules, retention and copies across services and accounts (with Organizations). Vault Lock in compliance mode makes the retention rules immutable once the cooling-off period ends: no one, including the root user, can shorten them or delete recovery points early. A separate account limits the impact of compromised credentials in the workload account. Test restores regularly too, because untested backups aren't evidence.
Why the others are wrong:
- A: Custom code still needs immutability and monitoring added. It's more work and more risk.
- C: Automated backups stay in the same account, and administrators can delete them.
- D: Exports with S3 replication cover only some data and need extra controls, such as Object Lock, to be immutable.
ECS Fargate tasks in private subnets with no NAT gateway fail to start with an error about pulling the container image from ECR. The task execution role has the correct ECR permissions. What fixes this while keeping the subnets private?
- A. Add only an interface VPC endpoint for
com.amazonaws.<region>.ecr.dkr - B. Add interface endpoints for
ecr.apiandecr.dkr, an S3 gateway endpoint (image layers are stored in S3), and a CloudWatch Logs interface endpoint if the tasks use theawslogsdriver - C. Give the tasks public IP addresses
- D. Add
ecr:*to the task role
Correct answer: B
Explanation: Pulling from ECR involves the ECR API (authentication and image manifests), the Docker registry endpoint, and S3 (image layers). Without a NAT gateway, each one needs its own endpoint. Log delivery and Secrets Manager or Parameter Store lookups at task start need endpoints too. Security groups on interface endpoints must allow HTTPS from the tasks.
Why the others are wrong:
- A:
ecr.dkron its own isn't enough. Withoutecr.apiand the S3 gateway endpoint, the pull still fails. - C: A public IP doesn't help in a subnet with no internet gateway route, and it goes against the private-by-default design.
- D: Image pulls use the task execution role, not the task role, and the execution role is already correct. The problem is the network.
A Security Hub check found an S3 bucket that allowed public read access. The bank wants to (1) prevent public access across the account and (2) detect and automatically fix any bucket that becomes non-compliant. Which two actions achieve this?
- A. Enable S3 Block Public Access at the account level, and enforce it with an SCP
- B. Deploy an AWS Config managed rule for public S3 buckets with automatic remediation using an SSM Automation runbook
- C. Turn on Amazon GuardDuty S3 Protection
- D. Review the Trusted Advisor dashboard every week
- E. Put CloudFront in front of every bucket
Correct answers: A and B
Explanation: Account-level Block Public Access overrides bucket policies and ACLs that would grant public access. An SCP that denies s3:PutAccountPublicAccessBlock stops anyone turning it off. Config with automatic remediation is the detective control that fixes drift on its own. Together they give defence in depth: prevention plus detection.
Why the others are wrong:
- C: GuardDuty S3 Protection detects suspicious data access. It doesn't remediate bucket configuration.
- D: Weekly manual review is slow and not automatic.
- E: CloudFront is for content delivery. It doesn't stop a bucket being made public.
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 Data Engineer – Associate (DEA-C01)
The data pipeline exam. It covers ingesting, transforming, storing, operating and governing data on AWS.
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.