The AWS Certified Security – Specialty (SCS-C03)

Network ACLs arestatelessand are evaluated in order based on rule number, with lower rule numbers taking precedence. According to AWS Certified Security – Specialty incident response guidance, network ACLs can be used toimmediately block traffic at the subnet levelwithout restarting instances or modifying their runtime state.

By adding a deny rule with alower priority number (10)that explicitly denies TCP traffic on port 22 from the offending IP address (10.0.1.5), the unauthorized SSH session is immediately interrupted. This approach satisfies the requirement to keep the instance running and to preserve memory for forensic analysis.

Option A violates the requirement because restarting the instance clears memory. Option C would disrupt all legitimate SSH access, not just the unauthorized session. Option D would block all internet access and could cause widespread service disruption.

AWS documentation emphasizes usingnetwork ACL deny rules for rapid, targeted containmentwhen immediate interruption is required without altering instance state.

AWS Certified Security – Specialty Official Study Guide

Amazon VPC Network ACL Documentation

AWS Incident Response Best Practices

AWS CloudFormationdynamic referencesprovide a secure mechanism for retrieving sensitive values from AWS Secrets Manager at stack creation or update time. According to the AWS Certified Security – Specialty documentation, dynamic references ensure that sensitive data such as database credentials arenever stored in plaintextin CloudFormation templates, parameters, stack metadata, or logs.

When a dynamic reference to Secrets Manager is used, CloudFormation retrieves the secret value at runtime and passes it securely to the resource that requires it. The secret value is not exposed to users who view the template, stack, or change sets.

Option B is insecure because parameters can be exposed through the CloudFormation console and APIs. Option C is incorrect because SecureString parameters are a feature of AWS Systems Manager Parameter Store, not Secrets Manager. Option D is invalid because KMS encrypts data but does not store secrets or manage secret rotation.

AWS best practices clearly state thatCloudFormation dynamic references to Secrets Managerare the recommended solution for securely handling sensitive configuration values.

AWS Certified Security – Specialty Official Study Guide

AWS CloudFormation Security Best Practices

AWS Secrets Manager Documentation

The immediate containment step for exposed access keys is todisable (deactivate) the compromised IAM access key(Option B). This prevents any further use of the leaked credentials, which is essential once secrets are publicly exposed. Creating a new key (Option D) may be part of recovery later, but it does not stop abuse of the already exposed key unless the exposed key is first deactivated.

To determine whether the credentials were used, you need evidence of access activity. Among the provided options, the best fit is generating and reviewing theIAM credential report(Option E). The report includes metadata such as access key status and “last used” style details that help triage whether the user’s credentials have been exercised recently. While deeper investigation would typically rely on CloudTrail “AccessKeyId” searches, the credential report is a quick AWS-native step aligned to the answer choices.

Option A is not correct: IAM Access Analyzer helps identify external access paths to resources and validate policies; it does not provide a definitive history of what a specific access key did. Option C is not a GuardDuty capability—GuardDuty generates findings; it does not “block” a specific access key. Therefore, deactivating the key and using credential reporting to assess recent usage best matches the requirements.

The requirement is to havekey material generated and used inside a custom key store backed by an AWS CloudHSM cluster. This is exactly whatAWS KMS Custom Key Storesprovide: KMS manages the keys and policies, but the cryptographic operations for those KMS keys occur in the associatedCloudHSMcluster, keeping the key material within HSM boundaries. For applications that needlocal-use data keys(both symmetric data keys and asymmetric data key pairs), KMS supports generating data keys and data key pairs that applications can use for envelope encryption and local cryptographic operations, while the master key protections remain within KMS (and within CloudHSM when using a custom key store).

For auditing, AWS best practice isAWS CloudTrail, which records KMS API calls (such as CreateKey, GenerateDataKey, GenerateDataKeyPair, Encrypt/Decrypt, etc.) and provides an immutable event history for compliance and investigation. Athena can query logs, but it is not the primary audit record source; GuardDuty is for threat detection, not authoritative key-usage auditing. Therefore, the correct combination isKMS with a CloudHSM-backed custom key storeplusCloudTrailfor auditability.

Amazon GuardDuty is afully managed, organization-aware threat detection servicethat continuously analyzes AWS logs such as CloudTrail events, VPC Flow Logs, DNS logs, EKS audit logs, and RDS activity. According to the AWS Certified Security – Specialty Official Study Guide, GuardDuty is designed to operate atscale across AWS Organizations with minimal operational overhead.

By designating a GuardDuty administrator account in the organization’s management account and enabling GuardDuty organization-wide, the company can automatically enable threat detection across hundreds of AWS accounts. EnablingEKS Protectionallows GuardDuty to analyze Kubernetes audit logs for suspicious activity, whileRDS Protectionprovides anomaly detection for Amazon Aurora databases.

Options B, C, and D require custom log aggregation, processing, and analytics pipelines, which significantly increase operational effort and maintenance complexity. Amazon Inspector does not analyze logs, Athena-based analysis is manual, and Kinesis plus Lambda requires custom detection logic.

AWS documentation explicitly identifiesGuardDuty with AWS Organizations integrationas the recommended solution for centralized, automated threat detection across multi-account environments with minimal operational effort.

AWS Certified Security – Specialty Official Study Guide

Amazon GuardDuty User Guide

GuardDuty Organization Administration Documentation

AWS incident response best practices emphasize rapid containment to prevent further data exposure. According to the AWS Certified Security – Specialty Study Guide, the fastest and least disruptive containment method for compromised compute resources is to immediately revoke credentials and permissions rather than modifying data or infrastructure.

Revoking the IAM role’s active sessions prevents the EC2 instance from continuing to access AWS services. Updating the S3 bucket policy to explicitly deny access to the IAM role ensures immediate enforcement, even if temporary credentials remain cached. Removing the IAM role from the instance profile further prevents new credentials from being issued.

Option A and D involve large-scale data movement or re-encryption, which is time-consuming and operationally expensive. Option B relies on network-level controls that do not prevent access through private AWS endpoints.

AWS guidance explicitly recommends credential revocation and policy-based denial as the fastest containment step during active incidents.

Referenced AWS Specialty Documents:

AWS Certified Security – Specialty Official Study Guide

AWS Incident Response Best Practices

AWS IAM Role Session Management

Option A satisfies all requirements with the most direct, purpose-built AWS logging workflow. By using the CloudWatch Agent (or fluent-bit / unified logging configuration) on each EC2 instance—regardless of whether it is On-Demand or Spot—the application logs can be centralized into asingle Amazon CloudWatch Logs log group. Centralization ensures the logs remain available even as Spot Instances are interrupted and replaced. Access control is handled withIAM policies(and optionally resource policies/KMS encryption) so that only a specific set of users can read/query the log group.

For analysis,CloudWatch Logs Insightsprovides an interactive query language that is SQL-like and commonly treated as “SQL queries” for troubleshooting. It enables fast filtering, aggregation, and pattern detection across large log volumes without building a separate data lake pipeline. This supports event-pattern analysis and root cause investigation directly from the centralized log group.

Option B is incorrect because Logs Insights queries CloudWatch Logs data, not arbitrary log files sitting in S3. Option C is inefficient (many log groups) and Athena cannot directly query CloudWatch log groups as a native data source. Option D is incorrect because Amazon Detective is for security investigations across supported data sources and is not the primary service for ad-hoc SQL-style querying of application logs.

A credential stuffing attack at the ALB is aLayer 7problem and is best mitigated withAWS WAF. The attacker is distributed across many IPs, so blocking by IP in a security group (Option B) is ineffective and operationally heavy. A CloudWatch alarm (Option A) only alerts; it does not block or mitigate requests.

Because the malicious traffic uses a distinctive, knownUser-Agentstring associated with a mobile device emulator, AWS WAF can quickly reduce the attack by inspecting the User-Agent header and blocking matching requests. This approach is targeted: it blocks the identified automated attack pattern while allowing legitimate users who do not present that emulator User-Agent to continue logging in. The WAF rule can be deployed immediately on the existing ALB-associated web ACL and can be further refined (for example, applied only to /login paths, combined with rate-based rules, or integrated with Bot Control) to minimize false positives.

Option D is risky because “allow only legitimate user agents” is brittle: user agents are diverse and change frequently, and a strict allow-list can accidentally block real users. Therefore, a WAF custom block rule for the known malicious User-Agent string is the correct solution.

To restrict activity to a single Region in an OU using an SCP, the standard pattern is an explicitDenyfor requests madeoutsidethe allowed Region, while carving out exceptions forglobal servicesthat do not use aws:RequestedRegion in the same way (or that must remain usable regardless of Region). This is done withEffect: Deny, aConditionusing StringNotEquals on aws:RequestedRegion for the allowed Region (here, eu-west-1), andNotActionlisting the global services that should remain available.

This works because SCPs act asguardrails: an explicit Deny in an SCP overrides IAM Allow in member accounts, ensuring the restriction applies consistently to all existing and future accounts placed in the OU. The StringNotEquals condition ensures the deny triggers for any Region other than eu-west-1. The NotAction exception list ensures that the specified global services are not blocked by this deny statement.

Option A is wrong because StringEquals would deny actionsineu-west-1 rather than outside it. Options B and D useAllowstatements, which do not enforce “only this Region” safely in SCPs unless combined with a comprehensive deny strategy; they would not reliably restrict all other services/regions. Therefore, option C is the correct SCP structure.