Skip to content

SAP-C03 · Domain 2: Security, Compliance, and Governance · 25% of the exam

Task 2.3: Design multi-account governance and compliance solutions.

Running an organization: OUs, Control Tower landing zones and account vending, central logging and configuration (AI model invocations included), resource sharing with AWS RAM, SCPs and RCPs, automated compliance checks, and vulnerability scanning and patching at scale.

Study it

  • Organizations, OUs, SCPs and RCPs, Control Tower and delegated administrators

    Partly covered by: Organizations, SCPs & Control Tower

  • Central logging, findings and compliance checks: organization trails, Config, Security Hub and GuardDuty

    Partly covered by: CloudWatch, Alarms & X-Ray

  • Resource sharing with AWS RAM, and vulnerability scanning and patching at scale

    Lesson coming

Sample questions

Try each one before opening the answer. Every option is explained, with the AWS documentation page that proves it.

Question 1 · choose 1

A company has 150 accounts in AWS Organizations and runs workloads in four Regions. The security team wants Amazon GuardDuty and AWS Security Hub CSPM turned on in every current and future member account, findings from every account and Region visible in one security tooling account, and high-severity findings sent to the on-call paging system through a single integration. What should a solutions architect do?

  1. ACreate an organization trail in AWS CloudTrail and an Amazon EventBridge rule in the logging account that matches GuardDuty and Security Hub CSPM findings recorded in the trail
  2. BMake the security account delegated administrator for both services with auto-enable, aggregate findings in one home Region, and add one EventBridge rule there
  3. CDeploy a CloudFormation StackSet that turns on both services in every account and creates an EventBridge rule and SNS topic in each account and Region for the pager
  4. DCreate an AWS Config aggregator in the security account for all accounts and Regions, and send its compliance change notifications for GuardDuty findings to the pager
Show the answer and why
  • ACreate an organization trail in AWS CloudTrail and an Amazon EventBridge rule in the logging account that matches GuardDuty and Security Hub CSPM findings recorded in the trail

    Incorrect

    CloudTrail records API activity in the accounts. Findings are produced by GuardDuty and Security Hub CSPM and are delivered as events, not collected through a trail.

  • BMake the security account delegated administrator for both services with auto-enable, aggregate findings in one home Region, and add one EventBridge rule there

    Correct

    A delegated administrator can manage GuardDuty and Security Hub CSPM for the organization's members. Cross-Region aggregation brings findings to the home Region, and Security Hub CSPM sends findings to EventBridge, where one rule can route them.

  • CDeploy a CloudFormation StackSet that turns on both services in every account and creates an EventBridge rule and SNS topic in each account and Region for the pager

    Incorrect

    This enables the services, but findings stay in each account and Region and the pager needs hundreds of separate subscriptions instead of one central integration.

  • DCreate an AWS Config aggregator in the security account for all accounts and Regions, and send its compliance change notifications for GuardDuty findings to the pager

    Incorrect

    A Config aggregator collects configuration and compliance data from AWS Config. It does not collect GuardDuty threat findings.

Central security operations combine three features: a delegated administrator per service with auto-enable for members, cross-Region aggregation to one home Region, and EventBridge rules on the aggregated findings.

Question 2 · choose 1

Auditors require that management events from every current and future account of an organization, in every Region, be delivered to one S3 bucket in a log archive account. Administrators of member accounts must not be able to stop or delete this logging, and auditors must be able to prove that log files were not changed after delivery. Which solution meets these requirements?

  1. ACreate a multi-Region organization trail with log file validation turned on that delivers to the bucket in the log archive account
  2. BDeploy a multi-Region trail to every account with a CloudFormation StackSet and point all trails to the bucket in the log archive account
  3. CCreate an AWS Config aggregator in the log archive account that collects configuration history from all accounts and Regions
  4. DStream each account's CloudTrail event history to Amazon CloudWatch Logs with a subscription filter to the log archive account
Show the answer and why
  • ACreate a multi-Region organization trail with log file validation turned on that delivers to the bucket in the log archive account

    Correct

    An organization trail logs events for all accounts in the organization, member accounts cannot modify or delete it, and log file validation lets auditors detect changes to delivered log files.

  • BDeploy a multi-Region trail to every account with a CloudFormation StackSet and point all trails to the bucket in the log archive account

    Incorrect

    A trail created in a member account belongs to that account, so its administrators can stop or delete it.

  • CCreate an AWS Config aggregator in the log archive account that collects configuration history from all accounts and Regions

    Incorrect

    AWS Config records resource configuration changes, not the API activity that CloudTrail management events capture.

  • DStream each account's CloudTrail event history to Amazon CloudWatch Logs with a subscription filter to the log archive account

    Incorrect

    Event history is per account and per Region and is not a trail that member administrators are prevented from changing; it does not give tamper-evident log files either.

An organization trail is created once, covers all accounts including new ones, and cannot be changed by members; log file validation provides the integrity proof.

Question 3 · choose 1

A networking team manages VPCs in a central network account. Twenty application teams, each with its own account in the same organization, must launch EC2 instances, RDS databases and Lambda functions into these centrally managed subnets. Application teams must not be able to change route tables, network ACLs or the VPCs themselves, and the company wants to keep the number of VPCs and transit gateway attachments low. What should a solutions architect recommend?

  1. AGive each team a VPC in its own account and attach every VPC to a shared transit gateway
  2. BPeer each application team's VPC with the central VPC and route the application subnets through it
  3. CShare the subnets of the central VPCs with the application accounts through AWS RAM
  4. DCreate a cross-account IAM role in the network account that lets the teams launch resources there
Show the answer and why
  • AGive each team a VPC in its own account and attach every VPC to a shared transit gateway

    Incorrect

    This works but multiplies VPCs and transit gateway attachments, and the application teams would own and be able to change their own route tables.

  • BPeer each application team's VPC with the central VPC and route the application subnets through it

    Incorrect

    Peering still gives every team its own VPC, which the company wants to avoid, and the teams would manage their own routing.

  • CShare the subnets of the central VPCs with the application accounts through AWS RAM

    Correct

    With VPC sharing, participant accounts create their own resources in subnets that the owner shares, while the owner keeps control of the VPC resources, which participants cannot modify.

  • DCreate a cross-account IAM role in the network account that lets the teams launch resources there

    Incorrect

    Resources would be created in the network account itself, so teams would lose separate ownership and billing, and the role would need broad permissions in the central account.

Centrally managed networks with workloads in separate accounts is VPC sharing: the owner shares subnets, participants launch into them, and only the owner changes the network.

Question 4 · choose 2

A security team needs two organization-wide preventive controls. First, no member account may leave the organization, whoever signs in. Second, S3 buckets in member accounts must never be readable by principals outside the organization, even if a bucket owner writes a bucket policy that grants such access. Which policies meet these requirements? (Choose TWO.)

  1. AAn SCP attached to the organization root that denies organizations:LeaveOrganization
  2. BA resource control policy that denies S3 access when the requesting principal is not in the organization, with exceptions for AWS service principals
  3. CAn SCP that denies s3:GetObject to every principal whose aws:PrincipalOrgID is not the organization's ID
  4. DA permissions boundary on every IAM role in the member accounts that denies organizations:LeaveOrganization
  5. EA tag policy that requires every S3 bucket to carry an owner tag with an approved organization value
Show the answer and why
  • AAn SCP attached to the organization root that denies organizations:LeaveOrganization

    Correct

    An SCP limits every principal in the member accounts it applies to, including the root user of those accounts, so a deny on LeaveOrganization stops any of them from leaving.

  • BA resource control policy that denies S3 access when the requesting principal is not in the organization, with exceptions for AWS service principals

    Correct

    Resource control policies set the maximum permissions on resources in member accounts, whoever the requesting principal is, so they can stop external access even when a bucket policy allows it.

  • CAn SCP that denies s3:GetObject to every principal whose aws:PrincipalOrgID is not the organization's ID

    Incorrect

    SCPs apply only to IAM users and roles in the member accounts. They have no effect on principals from outside the organization.

  • DA permissions boundary on every IAM role in the member accounts that denies organizations:LeaveOrganization

    Incorrect

    A permissions boundary limits only the IAM entities it is attached to. It does not apply to the root user, and administrators can create roles without it.

  • EA tag policy that requires every S3 bucket to carry an owner tag with an approved organization value

    Incorrect

    Tag policies standardize tags. They do not grant or deny access to a bucket.

SCPs cap what principals in your accounts can do; RCPs cap what can be done to resources in your accounts, including by outsiders. One requirement is about principals, the other about resources.

Question 5 · choose 1

A security team wants one data lake in its own account with security logs from AWS services across the organization plus logs from on-premises and SaaS tools, all normalized to one open schema so its analytics tools can query them together. Which service fits?

  1. AAmazon Security Lake
  2. BAWS Config aggregator in the security account
  3. CAmazon Inspector with a delegated administrator
  4. DCloudWatch Logs subscriptions in each account
Show the answer and why
  • AAmazon Security Lake

    Correct

    Security Lake centralizes security data from AWS, on-premises and SaaS sources into a data lake and converts it to the Open Cybersecurity Schema Framework.

  • BAWS Config aggregator in the security account

    Incorrect

    An aggregator collects configuration and compliance data, not security logs from many sources.

  • CAmazon Inspector with a delegated administrator

    Incorrect

    Inspector scans for vulnerabilities; it is not a log data lake.

  • DCloudWatch Logs subscriptions in each account

    Incorrect

    Subscriptions stream logs but do not normalize sources into one open schema in a data lake.

A normalized, central security data lake is Amazon Security Lake.

Question 6 · choose 1

AWS sends security notices for an organization's 150 member accounts to security contacts who left the company years ago. The central security team wants the security alternate contact of every member account set to its own distribution list. The team works only from its security tooling account, the management account is reserved for tasks that must run there, nobody may sign in as the root user of a member account, and the change must be scriptable so that it can be rerun as accounts are added. Which solution meets these requirements?

  1. ASubscribe the distribution list to the RSS feed of AWS Security Bulletins and retire the old addresses from the mail system
  2. BHave the root user of each member account sign in once and update the security contact on that account's Account page
  3. CConfigure AWS User Notifications in the management account to forward security events to the distribution list
  4. DDelegate AWS Account Management to the security tooling account and call PutAlternateContact for every member account
Show the answer and why
  • ASubscribe the distribution list to the RSS feed of AWS Security Bulletins and retire the old addresses from the mail system

    Incorrect

    Security Bulletins publish public posts about security issues that AWS investigates, and the feed keeps the posts of the last 30 days. They are not the notices that AWS sends about a specific account, which still go to the stale contacts.

  • BHave the root user of each member account sign in once and update the security contact on that account's Account page

    Incorrect

    A standalone account updates its own alternate contacts this way, and it would work here, but it means 150 root sign-ins, which the policy forbids, and it cannot be rerun as a script.

  • CConfigure AWS User Notifications in the management account to forward security events to the distribution list

    Incorrect

    User Notifications is a central location for managing AWS notifications, but it does not change the alternate contacts recorded on each account, and this setup would run from the management account.

  • DDelegate AWS Account Management to the security tooling account and call PutAlternateContact for every member account

    Correct

    Once trusted access for the Account Management service is turned on, a user in the management account or in a delegated admin account can update the alternate contacts of any member account through the console, the AWS CLI or the SDKs.

The constraints are central change, no management account in daily use, no root sign-ins and a repeatable script. The public bulletin feed and User Notifications keep the team informed but leave each account's contacts as they are. Root sign-ins would update the contacts but break the policy. A delegated admin for AWS Account Management can set the security alternate contact of every member account with PutAlternateContact from the security tooling account.

Question 7 · choose 1

A company will run 2,000 Amazon EC2 instances across 25 accounts in its organization. Its compliance standard requires critical operating system patches to be installed within seven days of release, during approved maintenance windows, with a central compliance report. Which solution meets these requirements with the LEAST effort?

  1. AUse a Systems Manager Patch Manager patch policy from Quick Setup for the organization, with baselines and a schedule
  2. BTurn on Amazon Inspector for the organization and open a ticket for each instance with an operating system vulnerability
  3. CAdd a cron job to the AMI of each instance that installs all available updates every night
  4. DDeploy an AWS Config managed rule that marks instances without the latest patches as noncompliant
Show the answer and why
  • AUse a Systems Manager Patch Manager patch policy from Quick Setup for the organization, with baselines and a schedule

    Correct

    A patch policy configured through Quick Setup applies patch baselines and schedules across accounts and Regions in an organization, and Patch Manager reports patch compliance for the managed nodes.

  • BTurn on Amazon Inspector for the organization and open a ticket for each instance with an operating system vulnerability

    Incorrect

    Inspector finds software vulnerabilities and reports them. It does not install patches, so people would still patch every instance.

  • CAdd a cron job to the AMI of each instance that installs all available updates every night

    Incorrect

    Unmanaged nightly updates ignore the approved windows and baselines and give no central compliance report.

  • DDeploy an AWS Config managed rule that marks instances without the latest patches as noncompliant

    Incorrect

    AWS Config evaluates and reports compliance, but a rule alone does not install patches on a schedule.

Patching on a schedule with baselines and an organization-wide compliance view is Patch Manager with patch policies.

Question 8 · choose 1

A company runs generative AI applications on Amazon Bedrock in 30 workload accounts of its organization, in two Regions each. Compliance requires the full prompt and response of every model invocation to be stored as objects in one S3 bucket in the log archive account, which only the security team can read. Which approach meets these requirements with the least custom code?

  1. AConfigure model invocation logging in every workload account and Region to deliver directly to the central bucket in the log archive account
  2. BDeploy invocation logging to a local bucket in each account and Region with StackSets, and copy the logs to the central bucket with S3 Replication
  3. CHave each application write its prompts and responses to Amazon SQS, and run a consumer in the log archive account that stores them in the bucket
  4. DTurn on model invocation logging only in the log archive account, because the logging configuration applies to the whole organization
Show the answer and why
  • AConfigure model invocation logging in every workload account and Region to deliver directly to the central bucket in the log archive account

    Incorrect

    Model invocation logging supports only destinations in the same account and Region as the logging configuration, so it cannot write straight to the log archive account's bucket.

  • BDeploy invocation logging to a local bucket in each account and Region with StackSets, and copy the logs to the central bucket with S3 Replication

    Correct

    Each account and Region logs to its own bucket as the service requires, StackSets rolls the configuration out across the organization, and S3 Replication copies the log objects into the central archive bucket.

  • CHave each application write its prompts and responses to Amazon SQS, and run a consumer in the log archive account that stores them in the bucket

    Incorrect

    This needs logging code in every application and a consumer service, while model invocation logging already captures full requests and responses without code.

  • DTurn on model invocation logging only in the log archive account, because the logging configuration applies to the whole organization

    Incorrect

    Invocation logging collects invocations made in the account and Region where it is configured. It does not cover other accounts of the organization.

Bedrock model invocation logging is configured per account and Region and writes only to destinations in that account and Region. Centralizing it means rolling out the configuration everywhere and then moving the logs, for example with S3 Replication.

Question 9 · choose 1

Engineers keep opening SSH (port 22) to 0.0.0.0/0 in security groups to debug instances, and the security team finds these rules days later. The company wants every such rule detected and removed within minutes of being created, with a record of each finding and fix. Which solution meets these requirements with the LEAST custom code?

  1. AReview the AWS Trusted Advisor security group check every week and delete the rules it lists
  2. BTurn on Amazon Inspector and review its network reachability findings for port 22 every day
  3. CTurn on CloudTrail Insights so that unusual security group changes raise an event for the security team
  4. DUse the restricted SSH AWS Config managed rule with automatic remediation by an Automation runbook
Show the answer and why
  • AReview the AWS Trusted Advisor security group check every week and delete the rules it lists

    Incorrect

    The check can list such rules, but a weekly manual review is neither automatic nor fast enough.

  • BTurn on Amazon Inspector and review its network reachability findings for port 22 every day

    Incorrect

    Inspector reports exposure findings, but a daily manual review does not remove the rule within minutes.

  • CTurn on CloudTrail Insights so that unusual security group changes raise an event for the security team

    Incorrect

    CloudTrail Insights detects unusual API call rates. It does not evaluate a rule's content or remove it.

  • DUse the restricted SSH AWS Config managed rule with automatic remediation by an Automation runbook

    Correct

    AWS Config evaluates security groups when they change, and its automatic remediation can run a Systems Manager Automation document against noncompliant resources, with the evaluation and remediation history kept by Config.

Detect-and-fix of a configuration within minutes is a Config rule with automatic remediation.

Practice domain 2 →Practice all domains →