# Free SAP-C02 practice questions

> 10 free SAP-C02 practice questions with the correct answer and an explanation.

Page: https://devopscodelab.com/exams/aws-sap-c02/free-practice-questions/
Practice exams: https://devopscodelab.com/exams/aws-sap-c02/

### Question 1 of 10

A company must back up every RDS database and EBS volume tagged backup=gold across 85 accounts daily, keep copies in a central backup account, and guarantee that even a compromised administrator in a member account or in the backup account cannot delete recovery points before 35 days. The backup configuration must be managed centrally. Which solution meets these requirements?

- A. Create Amazon Data Lifecycle Manager policies in each account that snapshot tagged volumes and share the snapshots with the backup account for copying
- B. Deploy AWS Backup plans to each account with CloudFormation StackSets and copy the recovery points to a versioned S3 bucket with MFA delete in the backup account
- C. Deploy the plan with Organizations backup policies, copying to a vault in the backup account, and lock that vault with Backup Vault Lock in compliance mode
- D. Enable cross-Region automated backups for every RDS instance, and protect EBS snapshots with Recycle Bin retention rules that keep them for 35 days

**Answer:** C

**Explanation:** Organizations backup policies centrally deploy backup plans to all accounts, the copy action puts recovery points in the central vault, and Vault Lock in compliance mode prevents anyone, including administrators and the root user, from deleting them before the retention period. StackSets work but recovery points cannot be copied into an S3 bucket by AWS Backup. Data Lifecycle Manager does not cover RDS, and cross-Region RDS backups and Recycle Bin rules are configured per account and can be changed by a compromised administrator.

### Question 2 of 10

A retail company connects production VPCs, development VPCs, a shared services VPC and an on-premises network through a single Transit Gateway. Production and development VPCs must never be able to route to each other. Both environments must reach the shared services VPC and on-premises networks, and new VPCs are added weekly by an automated pipeline. Which configuration meets these requirements with the LEAST operational risk?

- A. Create separate route tables per environment, associate each attachment with its environment's table, propagate shared services and on-premises routes into the production and development tables, and all VPC routes into the shared table
- B. Use the default Transit Gateway route table for every attachment and have the pipeline add static blackhole routes for each new development CIDR in production and each new production CIDR in development
- C. Replace the Transit Gateway with VPC peering from every VPC to the shared services VPC, and route development and production traffic to on-premises through appliances in the shared services VPC
- D. Keep the default Transit Gateway route table with propagation from every attachment, and have the pipeline add network ACL rules in every production and development subnet that deny the other environment's CIDR ranges

**Answer:** A

**Explanation:** Separate route tables per environment, with attachments associated to their environment's table and only shared services and on-premises routes propagated into them, make production and development unreachable from each other by design, and new attachments inherit the right behavior. Blackhole routes and NACL rules on a shared default table depend on the pipeline maintaining deny entries for every new CIDR, so one miss creates a path. VPC peering is not transitive, so the shared services VPC cannot act as a transit hub to on-premises without self-managed appliances.

### Question 3 of 10

A media company wants application teams in 50 accounts to deploy only pre-approved, encrypted RDS and ElastiCache configurations. Teams must not need permissions to create the underlying resources directly, and the platform team wants to publish new versions of the approved templates in one place for all accounts. Which solution meets these requirements?

- A. Publish the templates as CloudFormation modules in the private registry of each account and allow teams to use them inside their own stacks
- B. Deploy the approved databases into each requesting account with CloudFormation StackSets run by the platform team when a ticket is opened
- C. Publish AWS CDK constructs with encryption defaults to a shared AWS CodeArtifact repository and have teams deploy them from their own pipelines
- D. Create a Service Catalog portfolio with launch-constrained products and share it with the workload OUs through AWS Organizations

**Answer:** D

**Explanation:** Service Catalog portfolios shared through Organizations make approved products available in every account, launch constraints let products be provisioned through a role so users do not need direct resource permissions, and new product versions are published once. CloudFormation modules and shared CDK constructs both deploy with the team's own credentials, so teams still need permissions to create the underlying resources and can deviate from the approved configuration. Ticket-driven StackSets are not self-service.

### Question 4 of 10

A logistics company runs stateless microservices on Amazon EKS backed by an Aurora PostgreSQL database in us-east-1. All Kubernetes manifests are stored in Git and applied by a GitOps controller, and the infrastructure is defined in Terraform. The DR requirements are an RTO of 1 hour and an RPO of 5 minutes in us-west-2 at the lowest ongoing cost. Which design meets these requirements?

- A. Use AWS Elastic Disaster Recovery to continuously replicate the EKS worker nodes and the Aurora database hosts to a low-cost staging area in us-west-2, and launch recovery instances during a disaster
- B. Take Aurora cluster snapshots every 5 minutes with AWS Backup, copy each snapshot to us-west-2, and back up Kubernetes objects with Velero once a day so that the cluster can be restored there
- C. Run a second full-size EKS cluster in us-west-2 with every workload deployed and scaled to production capacity by the GitOps controller, backed by an Aurora global database secondary cluster
- D. Use an Aurora global database secondary in us-west-2 and ECR cross-Region replication, and create the EKS cluster and workloads there from Terraform and GitOps only during a disaster

**Answer:** D

**Explanation:** A pilot light keeps only the data layer replicating continuously, which meets a 5-minute RPO, while the EKS cluster and stateless workloads can be recreated from Terraform and GitOps well within an hour using replicated images. A full-size standby cluster meets the objectives but at much higher ongoing cost. Aurora snapshots cannot be taken and copied every 5 minutes reliably, so the RPO is at risk. Elastic Disaster Recovery replicates servers, not managed EKS control planes or Aurora clusters, which have no hosts to install its agent on.

### Question 5 of 10

A fintech company is building a balance-tracking service that must accept writes in three AWS Regions simultaneously. Every read must return the latest committed balance regardless of the Region it is served from, and a Regional outage must not cause the loss of any acknowledged write. The team wants a fully managed, serverless data store and is willing to accept higher write latency to achieve this. Which design meets these requirements?

- A. An Aurora PostgreSQL global database with write forwarding enabled on the secondary clusters
- B. A single-Region Amazon DynamoDB table with point-in-time recovery and AWS Backup cross-Region copies to the other two Regions
- C. An Amazon DynamoDB global table with the default multi-Region eventual consistency, using conditional writes on a version attribute
- D. An Amazon DynamoDB global table configured for multi-Region strong consistency, with strongly consistent reads in each Region

**Answer:** D

**Explanation:** DynamoDB global tables with multi-Region strong consistency replicate writes synchronously before acknowledging them, giving a zero RPO and strongly consistent reads in every participating Region at the cost of higher write latency. The default eventually consistent mode resolves conflicts with last-writer-wins and replicates asynchronously, so a Regional failure can lose recent writes and reads can be stale. Aurora write forwarding still relies on asynchronous storage replication to secondaries, and backup copies cannot provide multi-Region writes or a zero RPO.

### Question 6 of 10

A manufacturer streams sensor readings from factory equipment at roughly 50,000 events per second. Each event triggers a short validation and normalization workflow of four steps that completes in under two seconds. Steps are idempotent, so at-least-once execution is acceptable, and execution history is only needed in Amazon CloudWatch Logs. Which orchestration choice is the MOST cost-effective?

- A. AWS Step Functions Express Workflows started asynchronously, with logging to CloudWatch Logs
- B. Amazon Managed Workflows for Apache Airflow with one DAG run per event
- C. A chain of AWS Lambda functions connected through Amazon SQS standard queues, with each function invoking the next
- D. AWS Step Functions Standard Workflows with execution history retained by the service

**Answer:** A

**Explanation:** Express Workflows are priced on executions, duration, and memory, suit high-volume short-lived processing, and provide at-least-once semantics with history in CloudWatch Logs. Standard Workflows charge per state transition and provide exactly-once semantics and long-term history that are unnecessary here and cost far more at this volume. Airflow is designed for batch scheduling, not tens of thousands of runs per second, and hand-chained Lambda functions with queues add operational complexity and lose the visual orchestration.

### Question 7 of 10

During a routine CloudFormation stack update, a change to a property of a production Amazon RDS DB instance caused CloudFormation to replace the instance, and the original database was deleted. The company wants to prevent stack updates from ever silently replacing or deleting stateful resources, while still allowing other resources in the same stacks to be updated. Which combination of controls should be used?

- A. Require every stack update to pass a manual approval stage in AWS CodePipeline, where a senior engineer reviews the template diff in the pull request and signs off before CloudFormation applies the update to production
- B. Review change sets before execution, apply a stack policy denying Update:Replace and Update:Delete on stateful resources, and set DeletionPolicy and UpdateReplacePolicy to Snapshot or Retain on them
- C. Enable termination protection on every production stack and run drift detection before each update so that unexpected changes to the database are found before the update starts
- D. Deploy production stacks with CloudFormation StackSets using a maximum concurrency of one account so that a harmful update is caught in the first account before it reaches others

**Answer:** B

**Explanation:** Change sets reveal replacements before execution, a stack policy blocks replacement or deletion of protected resources during updates, and DeletionPolicy and UpdateReplacePolicy preserve data if a resource is removed anyway. Termination protection prevents stack deletion only, and drift detection does not stop replacements. StackSets change the rollout scope but still replace the database in the first account. A template diff does not show which property changes trigger replacement, so a manual approval without change sets does not reliably prevent the mistake.

### Question 8 of 10

A bank's network team must prove continuously that none of the subnets holding cardholder data in 30 VPCs can be reached from an internet gateway or from development VPCs, even as security groups, route tables and Transit Gateway attachments change. Today engineers check individual paths manually before each audit. Which solution meets this requirement with the LEAST effort?

- A. Run VPC Reachability Analyzer between each internet gateway and each instance in the cardholder subnets before every audit, and export the path results as audit evidence
- B. Define Network Access Analyzer scopes for prohibited paths from internet gateways and development VPCs to cardholder subnets, run analyses on a schedule, and alert on findings
- C. Use AWS Firewall Manager to apply a common security group policy to all cardholder data instances across the 30 VPCs, and audit for any security group rules that allow broader access
- D. Enable VPC Flow Logs on the cardholder subnets and alert through CloudWatch Logs metric filters when traffic from internet or development CIDR ranges is accepted

**Answer:** B

**Explanation:** Network Access Analyzer evaluates network configuration against scopes that describe disallowed access patterns across many resources and reports any path that exists, which suits continuous compliance. Reachability Analyzer checks one source-destination pair at a time, so covering all paths is laborious. Flow logs reveal traffic only after it happens, and a Firewall Manager security group policy controls one layer but does not prove that no route exists through route tables and Transit Gateway attachments.

### Question 9 of 10

An enterprise is migrating hundreds of internal applications that use TLS certificates issued by an on-premises Microsoft root CA trusted by all corporate devices. The security team wants certificates for migrated workloads issued and renewed automatically in AWS, without distributing a new root of trust to every device. Which approach meets these requirements?

- A. Request public certificates from AWS Certificate Manager for all internal hostnames and validate them with DNS records in a Route 53 private hosted zone
- B. Export certificates from the on-premises CA for each workload, import them into AWS Certificate Manager, and set an EventBridge reminder to reimport them before expiry
- C. Create a subordinate CA in AWS Private CA, have the on-premises root CA sign its certificate, and issue private certificates through ACM with managed renewal
- D. Create a new root CA in AWS Private CA and push its certificate to all corporate devices through Group Policy and the mobile device management platform

**Answer:** C

**Explanation:** A subordinate CA in AWS Private CA chained to the existing root is trusted by devices automatically, and ACM private certificates renew without manual work. A new root requires redistributing trust to every device. Public ACM certificates cannot be issued for internal-only hostnames, and validation records in a private hosted zone are not publicly resolvable. Imported certificates are not renewed automatically, so reminders still leave manual reimport work.

### Question 10 of 10

A semiconductor manufacturer is migrating most applications to an AWS Region, but its manufacturing execution system must respond within single-digit milliseconds to equipment controllers that physically remain on the factory floor. The company wants to use the same AWS APIs, tooling, and services such as Amazon EC2 and Amazon EBS for that system. Which solution meets these requirements?

- A. Deploy the system in the nearest AWS Local Zone and connect the factory to it over AWS Direct Connect
- B. Deploy the system on an AWS Outposts rack installed in the factory and connected to its parent Region
- C. Run the system on AWS Outposts servers on the factory floor and keep its data on instance store volumes
- D. Deploy the system in the parent Region and use AWS Global Accelerator to reduce latency from the factory

**Answer:** B

**Explanation:** An Outposts rack places AWS infrastructure on site, providing EC2 and EBS with the same APIs and local latency to the equipment. A Local Zone may be close but still adds network distance that cannot be guaranteed in single-digit milliseconds from the factory. Global Accelerator optimizes network paths but cannot overcome distance to the Region, and Outposts servers provide EC2 with instance storage but do not support Amazon EBS, which the system requires.

Ready for a timed run? Take the free 15-question SAP-C02 mini exam: https://devopscodelab.com/app/exams/aws-sap-c02/mini/

DevOpsCode Lab is not affiliated with AWS. SAP-C02 and related names are trademarks of their owners. We write every question ourselves; these are not real exam questions.
