Skip to content

SAP-C03 · Domain 4: Resilience, Migration, and Business Continuity · 15% of the exam

Task 4.3: Design solutions to meet business continuity and resiliency requirements.

Staying up and recoverable: backup governance across accounts, failover tests that measure recovery time, redundancy and self-healing designs, protection against ransomware and deletion with Object Lock, snapshot lock and Vault Lock, and controlled chaos experiments.

Study it

Sample questions

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

Question 1 · choose 2

A new web tier runs in three Availability Zones. The team wants it to keep serving full traffic if one zone fails, without having to launch instances or change configuration during the event. Which design choices support this? (Choose TWO.)

  1. ARun the fleet in a single zone and rely on Auto Scaling to launch elsewhere on failure
  2. BRun all of the capacity in each zone on Spot Instances to save cost
  3. CLaunch the instances in a cluster placement group for lower latency
  4. DPre-provision enough capacity in each zone that the remaining two can carry the full load
  5. EDesign recovery so it does not depend on control plane changes during the event
Show the answer and why
  • ARun the fleet in a single zone and rely on Auto Scaling to launch elsewhere on failure

    Incorrect

    Launching new instances during the failure depends on control plane actions and is the bimodal behavior to avoid.

  • BRun all of the capacity in each zone on Spot Instances to save cost

    Incorrect

    Spot Instances are returned when Amazon EC2 needs the capacity back, so the headroom the design relies on could be interrupted at any time, including during the zone event.

  • CLaunch the instances in a cluster placement group for lower latency

    Incorrect

    A cluster placement group keeps instances within a single Availability Zone for low-latency networking, which is the opposite of spreading capacity across zones.

  • DPre-provision enough capacity in each zone that the remaining two can carry the full load

    Correct

    Having the extra capacity pre-provisioned, instead of launching new instances during a zone failure, helps achieve static stability.

  • EDesign recovery so it does not depend on control plane changes during the event

    Correct

    Removing control plane dependencies from the recovery path helps produce more resilient, statically stable workloads.

Static stability means enough capacity is already in place and nothing has to change during a failure.

Question 2 · choose 2

A company claims that its order service survives the loss of an Availability Zone, but it has never tested this. Leadership wants a repeatable experiment in a production-like environment that simulates a zone outage and stops on its own if customer-facing error rates climb too high. Which actions meet these requirements? (Choose TWO.)

  1. ABuild an AWS Fault Injection Service experiment from the AZ Availability Power Interruption scenario
  2. BRun an AWS Resilience Hub assessment and treat its resiliency score as the result of the test
  3. CAsk an engineer to stop instances in one zone from the console during a quiet period
  4. DRun the experiment in a separate account that has no traffic and no copy of the service
  5. EAdd a stop condition to the experiment template that uses the CloudWatch alarm for the error rate
Show the answer and why
  • ABuild an AWS Fault Injection Service experiment from the AZ Availability Power Interruption scenario

    Correct

    This scenario induces the expected symptoms of a complete loss of power in one Availability Zone across the affected resource types, which is what the claim needs to be tested against.

  • BRun an AWS Resilience Hub assessment and treat its resiliency score as the result of the test

    Incorrect

    A Resilience Hub assessment evaluates the application against its resiliency policy. On its own, it does not inject a zone failure into the running system.

  • CAsk an engineer to stop instances in one zone from the console during a quiet period

    Incorrect

    A manual action is not repeatable, covers only some resource types and has no automatic stop if errors climb.

  • DRun the experiment in a separate account that has no traffic and no copy of the service

    Incorrect

    An experiment against an environment without the service and its traffic shows nothing about how the order service behaves.

  • EAdd a stop condition to the experiment template that uses the CloudWatch alarm for the error rate

    Correct

    Stop conditions use CloudWatch alarms to stop a running experiment when the alarm goes into the ALARM state.

Fault Injection Service gives a repeatable zone-outage scenario, and its stop conditions tie the blast radius to a CloudWatch alarm.

Question 3 · choose 1

After a ransomware attack at a competitor, a bank decides that the daily EBS snapshots of its core systems must be impossible to delete for 90 days, even by administrators or the root user of the account. The backup team also wants a short window right after each lock in which it can correct a lock that was applied with the wrong duration, before the lock becomes permanent. Which approach meets these requirements?

  1. ALock the snapshots in compliance mode for 90 days with a cooling-off period of a few hours
  2. BLock the snapshots in governance mode for 90 days and restrict the unlock permission to the backup team
  3. CCreate a Recycle Bin retention rule that keeps deleted snapshots for 90 days
  4. DLock the snapshots in compliance mode for 90 days without a cooling-off period
Show the answer and why
  • ALock the snapshots in compliance mode for 90 days with a cooling-off period of a few hours

    Correct

    A compliance-mode lock cannot be removed or shortened by anyone once the cooling-off period ends, and during the cooling-off period permitted users can still correct the lock settings.

  • BLock the snapshots in governance mode for 90 days and restrict the unlock permission to the backup team

    Incorrect

    In governance mode, users with the right IAM permissions can unlock the snapshot at any time, so an attacker with administrator access could remove the lock.

  • CCreate a Recycle Bin retention rule that keeps deleted snapshots for 90 days

    Incorrect

    Recycle Bin lets deleted snapshots be recovered, but it does not make them impossible to delete in the first place.

  • DLock the snapshots in compliance mode for 90 days without a cooling-off period

    Incorrect

    Without a cooling-off period, the lock takes effect immediately and can no longer be corrected if the duration was set wrong.

EBS snapshot lock offers governance mode, which privileged users can lift, and compliance mode, which nobody can lift. An optional cooling-off period of 1 to 72 hours gives a window to fix mistakes before a compliance lock becomes final.

Question 4 · choose 1

A payments application spans four AWS accounts in one Region: front end, orders, payments and shared data. The resilience team wants to run one controlled failure experiment that affects resources in all four accounts at the same time, start and stop it from one place, and keep a central log of every action taken in each account. Which approach meets these requirements?

  1. ACreate an experiment template in each account and start the four experiments at the same time from a script
  2. BCreate a multi-account AWS FIS experiment template in an orchestrator account, with a target account configuration per account
  3. CRun a Systems Manager Automation runbook in each account that stops instances, started at the same time by EventBridge Scheduler
  4. DWrite a Lambda function in each account that stops instances and blocks network traffic on a schedule
Show the answer and why
  • ACreate an experiment template in each account and start the four experiments at the same time from a script

    Incorrect

    Four separate experiments have no central control or shared log, and stopping one does not stop the others.

  • BCreate a multi-account AWS FIS experiment template in an orchestrator account, with a target account configuration per account

    Correct

    Multi-account experiments run from an orchestrator account that owns the template and experiment, act on target accounts through the roles in their target account configurations, and centralize logging.

  • CRun a Systems Manager Automation runbook in each account that stops instances, started at the same time by EventBridge Scheduler

    Incorrect

    Separate runbooks in four accounts give no single experiment to start and stop together, and no central log of the actions across accounts.

  • DWrite a Lambda function in each account that stops instances and blocks network traffic on a schedule

    Incorrect

    Custom scripts lack the experiment controls, stop conditions and central logging that the team wants.

AWS FIS multi-account experiments let one orchestrator account run a failure scenario across target accounts within a Region, with fine-grained roles per account and central visibility of the actions.

Question 5 · choose 1

A SaaS provider serves 5,000 tenants from one Multi-AZ stack in a single Region. Twice this year, one tenant's malformed requests or a bad configuration change took down the whole service for every tenant, although no Availability Zone failed. The provider wants a failure of this kind to affect only a small fraction of tenants in the future. Which architecture change meets this requirement?

  1. AAdd a third Availability Zone and raise the Auto Scaling maximum so that the stack has more spare capacity
  2. BRun the same stack active-active in a second Region behind latency-based routing
  3. CPartition tenants into isolated cells behind a thin routing layer and roll out changes one cell at a time
  4. DPut an SQS queue in front of the shared service so that malformed requests wait in the queue until it recovers
Show the answer and why
  • AAdd a third Availability Zone and raise the Auto Scaling maximum so that the stack has more spare capacity

    Incorrect

    More zones and capacity protect against infrastructure failures, but a poison request or bad change reaches every instance of the single stack.

  • BRun the same stack active-active in a second Region behind latency-based routing

    Incorrect

    Both Regions run the same code and configuration for all tenants, so a bad change or poison request can still affect everyone.

  • CPartition tenants into isolated cells behind a thin routing layer and roll out changes one cell at a time

    Correct

    In a cell-based architecture each cell is a complete, isolated copy of the workload for part of the traffic, so a failure stays within one cell and changes can be deployed to one cell at a time.

  • DPut an SQS queue in front of the shared service so that malformed requests wait in the queue until it recovers

    Incorrect

    A queue buffers load, but every request still flows to the same shared stack, so a poison request or bad change still affects all tenants.

Cell-based architecture reduces the scope of impact: tenants are split across isolated cells, so a software, configuration or poison-request failure affects only the tenants of one cell.

Question 6 · choose 1

Several teams run AWS Fault Injection Service experiments in a production account. During a real incident or a holiday change freeze, the operations lead must be able, in one action, to stop every experiment that is running in the account's Region and keep any new experiment from starting until the freeze ends. Which approach meets this requirement?

  1. AAttach an SCP that denies fis:StartExperiment to the production account for the duration of the freeze
  2. BDelete all experiment templates in the account when the freeze starts and re-create them afterward
  3. CRevoke the FIS experiment roles' permissions in IAM until the freeze ends
  4. DEngage the FIS safety lever for the account and Region, and disengage it when the freeze ends
Show the answer and why
  • AAttach an SCP that denies fis:StartExperiment to the production account for the duration of the freeze

    Incorrect

    Denying StartExperiment keeps new experiments from starting, but it does not stop the experiments that are already running.

  • BDelete all experiment templates in the account when the freeze starts and re-create them afterward

    Incorrect

    Deleting templates is destructive, slow to undo, and is not the mechanism FIS provides to halt running experiments.

  • CRevoke the FIS experiment roles' permissions in IAM until the freeze ends

    Incorrect

    Changing many roles is not one action, and it is not the documented way to stop running experiments and block new ones.

  • DEngage the FIS safety lever for the account and Region, and disengage it when the freeze ends

    Correct

    Engaging the safety lever stops in-progress experiments and cancels any experiment that is started while it is engaged, for that account and Region.

Each account has an FIS safety lever per Region. Engaging it halts all running experiments and prevents new ones until it is disengaged, which fits incidents and change freezes.

Practice domain 4 →Practice all domains →