DVA-C02 Sample Questions and the 720 Passing Score
The passing score first: 720 out of 1,000
The AWS Certified Developer – Associate (DVA-C02) exam is scored on a 100-to-1,000 scale, and the passing score is 720. The exam itself is 65 questions in 130 minutes — two minutes per question on average. Your result is a scaled score, not a raw percentage, and AWS reports your performance by domain.
One framing point before the samples: DVA-C02 is a developer exam, not a smaller version of Solutions Architect Associate. It cares less about whiteboard architecture and more about how applications are built, deployed, secured, and debugged on AWS — Lambda behaviour, DynamoDB design decisions, CI/CD mechanics, IAM evaluation, and reading telemetry when something is slow. The five original questions below are written in that spirit. They are practice items in the exam's style, not real exam questions.
Five worked sample questions
Sample question 1. A Lambda function sits behind an Amazon API Gateway REST API. Clients report that repeated identical GET requests are slow, and the function's logs show it recomputing the same result on every call. The underlying data changes only a few times per day. What is the MOST effective first step?
- A. Increase the Lambda function's memory allocation
- B. Enable API Gateway caching on the method with a suitable TTL
- C. Move the function to a provisioned concurrency configuration
- D. Place a CloudFront distribution in front of the API
Correct answer: B. If identical requests return identical results for hours, the cheapest fix is to stop invoking the function for them: API Gateway caching serves repeat responses from the cache until the TTL expires. Option A treats a compute problem that doesn't exist — the function isn't slow, it's being called unnecessarily. Provisioned concurrency (C) addresses cold starts, not redundant invocations. CloudFront (D) is a heavier tool for a problem that caching at the API already solves directly.
Sample question 2. A DynamoDB table stores user profiles keyed by user ID. A new feature must frequently find all profiles for a given email address, and the access pattern needs the most up-to-date data because it drives a security check. What should the developer do?
- A. Perform a Scan with a filter expression on the email attribute
- B. Create a global secondary index on the email attribute and query it with strongly consistent reads
- C. Create a local secondary index on the email attribute
- D. Enable DynamoDB Streams and copy the data to a second table keyed by email
Correct answer: D. The consistency requirement decides this one. A global secondary index can be created on an existing table at any time, but GSI queries support eventual consistency only — a GSI can never be read with strongly consistent reads, so option B fails on its own terms. A local secondary index (C) must be created with the table and cannot be added afterwards. A Scan (A) reads the whole table and filters afterwards — the classic wrong answer at scale. Copying the data through DynamoDB Streams into a second table keyed by email (D) makes email a key attribute, so the security check can query that table with strongly consistent reads and always see the most up-to-date data.
Sample question 3. A team deploys a Lambda application with AWS SAM. They want new versions to shift traffic gradually and roll back automatically if error rates spike. Which configuration achieves this with the least custom code?
- A. Deploy with CodeDeploy using a linear deployment preference in the SAM template and a CloudWatch alarm as the rollback trigger
- B. Manually update the production alias weights after each deployment
- C. Run two stacks and switch Route 53 records when satisfied
- D. Use a blue/green CodePipeline stage that swaps Lambda function ARNs
Correct answer: A. SAM integrates directly with CodeDeploy for Lambda: a deployment preference such as a linear or canary shift, declared in the template, moves traffic between versions, and a CloudWatch alarm can halt and roll back the deployment automatically. Option B is the manual version of the same idea — it works once and fails process-wise forever. Option C works for some workloads but is far heavier than SAM's native mechanism. Option D misdescribes how Lambda deployments work — versions and aliases, not swapped function ARNs, are the mechanism.
Sample question 4. An application needs database credentials that must be rotated automatically every 30 days, with previous versions retained for rollback. Which storage choice fits best?
- A. AWS Systems Manager Parameter Store standard parameters
- B. AWS Secrets Manager
- C. Environment variables on the Lambda function, encrypted with a KMS key
- D. An encrypted S3 object read at startup
Correct answer: B. Automatic, scheduled rotation with version history is Secrets Manager's defining feature — it integrates with RDS and other services to rotate credentials natively. Parameter Store (A) is the strongest distractor: it stores configuration well, but it has no built-in rotation engine, so the developer would have to build that logic. Environment variables (C) require a redeploy to change and offer no rotation workflow. An S3 object (D) is storage, not a secrets system — no rotation, no fine-grained audit trail.
Sample question 5. Users report intermittent latency spikes in an application made of several Lambda functions calling DynamoDB. The team needs to find which segment of the request path is slow before changing anything. What should they do first?
- A. Increase DynamoDB provisioned capacity and observe
- B. Enable AWS X-Ray tracing and inspect the service map and traces for the slow segment
- C. Raise the Lambda timeout values
- D. Turn on detailed CloudWatch metrics for the API
Correct answer: B. "Which part is slow?" is a tracing question, and X-Ray exists precisely to answer it: the service map shows where time accumulates across functions and downstream calls, and individual traces show the slow segment. Options A and C change the system before diagnosing it — capacity and timeouts may hide the symptom while the cause survives. Detailed metrics (D) tell you that something is slow, but not how time is distributed across a request path. Diagnose first, then fix the segment the trace indicts.
Pacing: the two-minute discipline
Sixty-five questions in 130 minutes is comfortable only if you protect the average. Work at roughly two minutes per question, flag anything that stalls you, and move on — a flagged question revisited with fresh eyes beats five minutes burned on one item. Developer questions often hinge on one clause ("strongly consistent", "least custom code", "first step"), so read the requirement sentence twice before reading the options.
One wrinkle: not every question is scored
Like other AWS exams, DVA-C02 can include a small number of unscored pretest questions that AWS is evaluating for future use. You cannot tell which ones they are, and they do not count toward your 720. The practical consequence: never let one bizarre, oddly-worded question rattle you or eat your clock. It may literally not matter. Answer it, flag it if you like, and move on.
Practise diagnostically, by domain
The most common weak preparation pattern is grinding mixed question sets and watching one overall score. Your AWS score report breaks performance down by domain, so your practice should too: track your accuracy separately for development, security, deployment, and troubleshooting topics, and let the weakest domain set your next study block. Five questions done carefully — reading why the best distractor fails — teach more than fifty done on autopilot. Timed, domain-scored free practice tests on Study4Pass are one way to run that diagnostic before booking the real sitting. Beforehand, it also helps to review the DVA-C02 roadmap and retirement outlook, the DVA-C02 retake policy, and whether the exam includes live labs or performance-based tasks.