AWS Certified CloudOps Engineer - Associate (SOA-C03) #1 Exam Introduction — Exam Structure and Study Roadmap
If you’ve run infrastructure directly on the console and CLI through the 27-part AWS practical track, organized the design perspective with SAA-C03, and the development perspective with DVA-C02, the next step is taking that infrastructure to the point where you can operate it and respond to incidents. AWS certifications come in four tiers — Foundational, Associate, Professional, and Specialty — and the Associate exam that operations roles aim for is exactly AWS Certified CloudOps Engineer - Associate (SOA-C03).
This post is the starting point of the series. Before you sit down for the exam, you need a mental picture of what the exam asks, how it asks it, what’s different from SAA and DVA, and how to prepare so you clear 720 points within 130 minutes. That’s what this post organizes.
First, the name: from SysOps to CloudOps #
When you search for this exam, the old name AWS Certified SysOps Administrator - Associate also comes up. They aren’t two separate exams — they are two generations of the same certification.
- SOA-C02 (SysOps Administrator): its last testing day was September 29, 2025, so it has already retired.
- SOA-C03 (CloudOps Engineer): the current exam, in effect since September 30, 2025. The name changed from SysOps Administrator to CloudOps Engineer.
So as of 2026, the only exam you can sit is SOA-C03. Plenty of materials still on the market are based on SOA-C02, so always check that the exam code on your material is C03 — that’s the first pitfall to avoid. What changed in the move from C02 to C03 is covered in a separate section below.
What kind of certification is SOA-C03? #
SOA-C03 asks whether you can monitor an already-deployed AWS environment, automate it, respond to incidents, and manage cost and performance. Where SAA asked “what do you design and how?”, the CloudOps Engineer asks about the operations that happen after that design is running.
Essentially every question on the exam is a variation of a single question.
“When this kind of symptom (an alarm, a failure, performance degradation, a security event) occurs in this environment, what is the most appropriate operational action?”
So rote memorization won’t raise your score. Even for the same service, you have to be able to distinguish, against operational scenarios, when a CloudWatch Alarm is the answer and when an EventBridge rule is, and when Auto Scaling is the answer and when Systems Manager Automation is. Someone who passes this exam is someone who can use the operational flow of detecting failures and automating recovery in an AWS environment as a yardstick for design judgment.
Who finds it valuable? #
| Role | Benefit |
|---|---|
| Infrastructure / DevOps engineer | Proves monitoring, automation, and incident response as official competencies. The core Associate for operations roles |
| SRE / platform engineer | Gains a shared vocabulary for reliability and continuity design and operations automation |
| Backend / full-stack developer | Becomes able to directly own the operational side (logs, metrics, recovery) of their own application |
| Cloud operations beginner | Makes SAA’s design knowledge one step more concrete, in an “operable” form |
Among the four Associate certifications (CLF is Foundational), SOA-C03 is the exam operations roles most often choose after SAA. Think of it as the certification that separates people who only know design from people who also know operations.
What’s different from SAA and DVA? #
If you’ve already gone through the SAA-C03 series or the DVA-C02 series, pinning down how the three exams differ in angle helps set your study direction.
| Aspect | SAA-C03 (design) | DVA-C02 (development) | SOA-C03 (operations) |
|---|---|---|---|
| What it asks | Choosing an architecture that fits requirements | How to handle AWS services in code | Operating a deployed environment and responding to incidents |
| Representative topics | VPC design, DB choice, DR patterns | Lambda, API Gateway, DynamoDB SDK | CloudWatch, Systems Manager, automated recovery |
| Point in time | Before building | While building | After building |
| Question character | Design trade-offs | Code/SDK behavior | Operational scenarios, troubleshooting |
The three exams overlap considerably in their service lists. What differs is the point in time they ask about. Where SAA asks “is Aurora or DynamoDB right for this workload?”, SOA asks “the CPU on a running RDS instance stays at 100% — how do you diagnose and act on it?” For the same RDS, it asks about operational response, not design choice.
Exam structure #
The format of SOA-C03 is as follows.
- Number of questions: 65 (of which 50 scored + 15 unscored)
- Exam time: 130 minutes
- Passing score: 720 / 1000 (scaled score)
- Exam fee: 150 USD
- Question format: Multiple choice (4 options, 1 correct) + Multiple Response (5+ options, 2+ correct)
- Validity: 3 years
- How to take it: testing center or online proctoring (OnVUE)
- Languages: offered in multiple languages, including Korean
In the early SOA-C02 days there was a hands-on exam (Exam Lab) where you performed tasks directly in the console, but it was removed in March 2023 and has not returned in SOA-C03. So the current exam consists entirely of multiple choice and Multiple Response. The 15 unscored questions are ones AWS mixes in for future use; they don’t count toward your score, and since you can’t tell which ones are unscored, just solve every question the same way. Against the 50 scored questions, a 720 means the passing line is roughly 70% correct or so.
The five domains #
SOA-C03 is divided into five domains, with the weights below. These weights are, in effect, the basis for allocating your study time.
| Domain | Weight | Core question |
|---|---|---|
| Domain 1. Monitoring, Logging, Analysis, Recovery, and Performance | 22% | How do you detect what happened and recover automatically? |
| Domain 2. Reliability and Business Continuity | 22% | How do you keep service running and data preserved even when failures occur? |
| Domain 3. Deployment, Provisioning, and Automation | 22% | How do you deploy environments consistently as code and automate repetitive work? |
| Domain 4. Networking and Content Delivery | 18% | When connectivity breaks, where do you look and how do you fix it? |
| Domain 5. Security and Compliance | 16% | How do you track who did what and enforce compliance? |
The first three domains (Monitoring, Reliability, Deployment Automation) at 22% each make up about two thirds of the exam. These three are the center of the CloudOps Engineer, which is why this series assigns eight posts, from #2 through #9, to these three domains. CloudWatch, Systems Manager, and CloudFormation recur throughout the exam, so it’s most efficient to nail them down the most solidly.
What changed from SOA-C02 to C03 #
For people studying with C02-based material, let me note what was added. By AWS’s own announcement, no topics were dropped from the existing scope; new topics were added and the exam was reorganized.
- Container operations folded in: ECS, EKS, and ECR entered the exam scope. Beyond just launching containers, they’re covered from a logging, monitoring, and deployment operations perspective.
- IaC strengthened: the scope, once centered on CloudFormation, widened to include CDK, Terraform, and Git.
- Multi-account emphasis: the weight of AWS Organizations and multi-region operations grew.
- Domain reorganization: C02’s six domains were merged into five, and performance optimization moved out of a separate domain and into the Monitoring domain.
The point is that the scope of operations widened from EC2-centric systems to include containers and multiple accounts. Container operations are covered separately in #9 of this series.
Exam format: how to read operational scenario questions #
SOA-C03 questions generally follow this structure.
- Situation description. “A company operates a certain workload, and a certain symptom has occurred.”
- Constraints. “Recover automatically while minimizing operational overhead.”
- Question. “Which of the following is the most appropriate action that meets the requirements?”
What separates scores here is the keyword in the constraint. For the same symptom, the correct answer changes depending on whether the condition is “least operational overhead,” “automated recovery without manual intervention,” or “fastest diagnosis.” For example, if the condition is “without manual intervention,” an automated recovery answer that ties EventBridge to Systems Manager Automation is more likely correct than an answer where a person acts in the console.
When reading a question, the habit of reading the constraints and the question before the situation description, then filtering the options by that standard, saves a great deal of time.
Study strategy #
This series is designed on the following premises.
- Layer operational vocabulary on top of hands-on feel. If you’ve handled services directly through the 27-part AWS practical track, the exam is the process of organizing that experience into “operational judgment criteria.” Without hands-on experience it’s hard to grasp the context of the questions from the exam alone, so I recommend running the practical track in parallel.
- Assume SAA’s design vocabulary. The VPC, DB, and DR design covered in the SAA-C03 series is used as-is in SOA, and SOA adds an operational layer on top. If you haven’t gone through SAA, I recommend running it in parallel.
- Allocate time by domain weight. Spending the most time on the first three domains (22% each) and the least on Security (16%) gives the best score efficiency.
- Concentrate mock exams in the later stage. After finishing the domain study (#2–#13), wrap up with #14 exam tips and #15 mock exam. A mock exam isn’t about checking knowledge — it’s a tool for practicing time allocation and identifying pitfalls.
65 questions in 130 minutes is an average of 2 minutes per question. Operational scenarios are long, so time per question is tight; it’s important to get in the habit of marking unknown questions (Mark for Review) and moving on, then revisiting them at the end.
Common Pitfalls #
1) Studying with SOA-C02 material #
The most common pitfall. A good share of courses and question banks on the market are still C02-based, so they leave out the expanded container and IaC scope or describe the hands-on exam that no longer exists. First check that your material’s exam code is C03.
2) Mistaking them for design questions #
People who took SAA first make the mistake of reading operational questions like design questions. SOA asks not “what do I choose and build?” but “what do I act on in an already-running environment?” You have to judge the options by operational appropriateness, not design excellence.
3) Skimming past the keyword in the constraint #
Expressions like “least operational overhead,” “without manual intervention,” and “fastest recovery” decide the correct answer. If you focus on the symptom description and miss the condition in the last line, you’ll pick the wrong option.
4) Missing the number of correct answers in Multiple Response questions #
If you miss the “Choose TWO” or “Choose THREE” notation and pick only one, it’s an automatic miss. You need the habit of always checking the count of correct answers in the last line of the question.
Summary #
What we covered in this post:
- SOA-C03 verifies your ability to operate a deployed AWS environment — an Associate certification. It was renamed from SysOps Administrator to CloudOps Engineer, and the old SOA-C02 retired on 2025-09-29
- 65 questions / 130 minutes / 720 points / 150 USD / 3-year validity. The hands-on exam was removed. Against the 50 scored questions, roughly 70% correct is the passing line
- Five domains. Monitoring (22%) , Reliability (22%) , Deployment Automation (22%) , Networking (18%) , Security (16%). The first three domains are two thirds of the exam
- Differences from C02. Container (ECS, EKS, ECR) folded in, IaC (CDK, Terraform) strengthened, Organizations and multi-region emphasized
- Study strategy. Layer the operations perspective on top of the practical track and SAA vocabulary. Allocate time by domain weight. Concentrate mock exams in the later stage
- Pitfalls. Studying with C02 material, mistaking them for design questions, skimming past constraint keywords, missing the number of correct answers in Multiple Response
Next: Domain 1-1 CloudWatch Metrics and Alarms #
We’ve organized the exam structure. Now we move into the first topic of the highest-weight Monitoring domain.
In #2 Domain 1-1 Monitoring — CloudWatch Metrics, Alarms, and Dashboards, I’ll organize how CloudWatch collects metrics, the difference between standard and custom metrics, alarm states and threshold design, Composite Alarms, and a dashboard setup that shows operational status at a glance.