SaveMyCert
Log in
5 of 5 free questions left today·for 10 a day
SOA-C03 · Domain 2

Reliability and Business Continuity practice questions

Reliability and Business Continuity is worth 22% of the SOA-C03 exam — the heaviest of the 5 domains. Scalability and elasticity, highly available and resilient environments, and backup/restore strategies that meet RTO and RPO. 6 fully worked examples are further down this page, answers included.

Exam weight
22%
the heaviest of the 5 domains
Questions
114
across 3 topics
Free, no account
5/day
a free account allows 10 a day
Explanations
Every option
right and wrong

Build a practice session

5 free questions left today.

Domains

How many?

5 free questions left today.

Mode

Ready when you are

10 fresh questions drawn across 1 of 5 domains, in Learn mode.

Review what you missed

Every question you answer incorrectly, and every question you flag while practising, is saved here automatically. Finish a session and you can come back to re-drill just those.

6 sample Reliability and Business Continuity questions, fully explained

Questions from the SOA-C03 bank mapped to domain 2, with the answer key and the reasoning behind every option. None of them repeat the examples on the main SOA-C03 practice page.

Question 1Reliability and Business Continuity

An operations team wants an Auto Scaling group to add one instance when average CPU exceeds 70% but add four instances at once when average CPU exceeds 90%, using one policy attached to one CloudWatch alarm. Which scaling policy type can respond this way?

Choose one.

  • a
    Target tracking scaling

    Target tracking takes a single target value and manages its own alarms; it does not let you define different adjustment sizes for different breach severities.

  • b
    Step scaling Correct

    Step scaling defines step adjustments tied to alarm breach magnitude — +1 instance for the 70–90% band and +4 above 90% — so the response scales with severity.

  • c
    Simple scaling

    Simple scaling ties one alarm to one fixed adjustment and then waits out a cooldown, so it cannot respond differently to a small breach versus a large one.

  • d
    Scheduled scaling

    Scheduled scaling changes capacity at specific times on a schedule; it does not react to CPU at all.

The concept

Step scaling sizes the capacity adjustment to how far the metric breached the alarm threshold, using step adjustments you define.

Why that’s the answer

Only step scaling lets you say 'a small breach adds one instance, a large breach adds four' — the adjustment scales with breach magnitude. Target tracking holds one value and hides its alarms; simple scaling makes exactly one fixed adjustment per cooldown; scheduled scaling ignores metrics entirely.

How to reason it out
  1. Identify the requirement: different response sizes at different severity levels of the same metric.
  2. Create a CloudWatch alarm on average CPU for the Auto Scaling group.
  3. Attach a step scaling policy with step adjustments: +1 instance for the 70–90% band, +4 instances above 90%.
  4. Configure instance warmup so newly launched instances are counted correctly during multi-step responses.

Exam tip: Adjustments sized to how far the alarm is breached are the signature use case for step scaling.

Auto Scaling and Elasticity: EC2, Caching, and Database Scaling — the lesson that teaches this.

Question 2Reliability and Business Continuity

A payroll application's EC2 fleet is overwhelmed every weekday at 9:00 AM when employees log in, and traffic falls off sharply after 6:00 PM. The pattern is identical every business day. Which scaling approach puts the extra capacity in place before the morning surge arrives?

Choose one.

  • a
    A simple scaling policy that adds instances when a CPU alarm fires

    A CPU alarm fires only after the surge has already started, and new instances then need time to boot — users hit the overload every morning.

  • b
    A higher maximum capacity on the Auto Scaling group for weekdays

    Maximum capacity is only a ceiling. Raising it launches nothing; some policy still has to raise desired capacity, and a reactive one would still be late.

  • c
    Scheduled actions that raise desired capacity at 8:30 AM on weekdays Correct

    Correct. A recurring scheduled action raises capacity before the known 9:00 AM surge, and a second action lowers it after 6:00 PM, so capacity follows the fixed business-day pattern.

  • d
    A shorter health check grace period so instances enter service faster

    The grace period only delays health evaluation; shortening it does not launch anything earlier and risks replacing instances that are still starting.

The concept

Scheduled scaling changes an Auto Scaling group's min, desired, and max capacity at specific times or on a recurring schedule, which suits predictable demand cycles.

Why that’s the answer

The load pattern is known and repeats every business day, so capacity should be provisioned ahead of time rather than in reaction to a breaching metric. Reactive policies (simple, step, target tracking) all launch after load has risen; raising the max alone launches nothing; the grace period is a health-evaluation setting, not a capacity trigger.

How to reason it out
  1. Confirm the demand pattern is time-driven and recurring, not random.
  2. Create a recurring scheduled action that raises min and desired capacity shortly before 9:00 AM on weekdays.
  3. Create a second recurring action that lowers capacity after 6:00 PM.
  4. Optionally keep a target tracking policy alongside for unexpected intraday spikes.

Exam tip: Predictable, clock-driven demand cycles are the trigger phrase for scheduled scaling — provision before the surge, not after.

Auto Scaling and Elasticity: EC2, Caching, and Database Scaling — the lesson that teaches this.

Question 3Reliability and Business Continuity

An e-commerce fleet follows a daily traffic curve, but the time of the peak drifts by an hour or more from day to day. Application servers need nearly 15 minutes to bootstrap, so reactive policies add capacity too late. The team does not want to maintain a calendar of capacity changes. Which policy type should the engineer add to the Auto Scaling group?

Choose one.

  • a
    Target tracking scaling with a lower average CPU target value

    A lower target keeps more spare capacity all day, which costs money, and it still reacts only after load arrives — the 15-minute bootstrap remains uncovered.

  • b
    Scheduled scaling with an action set for each weekday's peak

    Scheduled actions would have to be maintained by hand, which the team rejects, and fixed times miss a peak that drifts by an hour or more.

  • c
    Step scaling with larger adjustments at lower alarm thresholds

    Larger, earlier steps are still reactive: the alarm must breach before anything launches, and the bootstrap delay still leaves users waiting.

  • d
    Predictive scaling based on the group's recorded load history Correct

    Correct. Predictive scaling forecasts the recurring curve from the group's metric history and launches capacity ahead of the forecast, with a scheduling buffer to cover long boot times — no calendar to maintain.

The concept

Predictive scaling analyses an Auto Scaling group's historical load, forecasts the coming days and schedules capacity ahead of the forecast — designed for recurring patterns and instances that take a long time to become ready.

Why that’s the answer

Two constraints decide it: instances need 15 minutes of lead time (so any reactive policy is late) and the team will not maintain a schedule (which rules out scheduled actions, and fixed times also miss a drifting peak). Predictive scaling learns the curve from history and launches ahead of it; it is usually paired with target tracking to catch whatever the forecast misses.

How to reason it out
  1. Note why reactive scaling fails: the bootstrap time exceeds how fast an alarm-driven launch can help.
  2. Rule out scheduled actions: the team will not maintain them and the peak time drifts.
  3. Choose predictive scaling, which needs recurring load history and forecasts capacity ahead of demand.
  4. Optionally start in forecast-only mode to compare the forecast with actual load, then switch to forecast and scale.

Exam tip: Recurring load plus long instance start-up and no appetite for schedules → predictive scaling (alongside target tracking).

Auto Scaling and Elasticity: EC2, Caching, and Database Scaling — the lesson that teaches this.

Question 4Reliability and Business Continuity

After a new deployment, an Auto Scaling group that uses ELB health checks terminates every newly launched instance about 90 seconds after launch — before the Java application finishes its three-minute startup. Instances that survive serve traffic normally. The engineer must fix this without slowing how quickly instances that fail later in service are detected. What should the engineer change?

Choose one.

  • a
    Raise the group's health check grace period above the startup time Correct

    Correct. The grace period only applies to newly launched instances: the group ignores health check results until it expires, so a value above the startup time stops the premature replacements while in-service checks stay as fast as before.

  • b
    Raise the target group's unhealthy threshold count to cover startup

    A higher unhealthy threshold would also let instances survive startup, but it applies to every target all the time, so failures in service would take longer to detect — exactly what the engineer must avoid.

  • c
    Turn on scale-in protection for instances while they are starting

    Scale-in protection only stops an instance from being chosen during scale-in. It does not stop the group from replacing an instance that it considers unhealthy.

  • d
    Shorten the target group's health check interval so starts show sooner

    A shorter interval makes the load balancer reach its unhealthy verdict even faster, so instances would be killed earlier in their startup, not later.

The concept

The Auto Scaling health check grace period delays health evaluation for newly launched instances only, giving the application time to start before ELB health check failures can trigger a replacement.

Why that’s the answer

Instances die at 90 seconds while the app needs three minutes, so the group is acting on ELB failures before startup finishes. Raising the grace period above the startup time fixes only the launch window. Raising the target group's unhealthy threshold also hides the startup failures but slows detection for every in-service target, which the stem rules out; scale-in protection does not apply to health replacements.

How to reason it out
  1. Match the symptom: replacements happen at a fixed time after launch, before the application is ready.
  2. Recognise that the group starts acting on ELB health results once the grace period ends.
  3. Set the grace period above the measured startup time (here, above three minutes).
  4. Leave the target group thresholds alone so in-service failures are still caught quickly.

Exam tip: New instances replaced before they finish starting → raise the health check grace period, not the load balancer thresholds.

Auto Scaling and Elasticity: EC2, Caching, and Database Scaling — the lesson that teaches this.

Question 5Reliability and Business Continuity

An Auto Scaling group runs worker instances that render video jobs lasting up to two hours. When the group scales in during quiet periods, it sometimes picks a busy instance, and the job restarts on another worker. The team wants scale-in to remove idle workers instead of waiting for busy ones to finish. What should the engineer configure?

Choose one.

  • a
    A termination lifecycle hook that each worker completes after its job

    A Terminating:Wait hook lets a busy worker finish before it goes, but the busy instance has still been chosen — idle workers stay running while scale-in waits up to two hours.

  • b
    The OldestInstance termination policy on the Auto Scaling group

    Termination policies choose by age, launch template or allocation strategy; none of them knows whether an instance is mid-job, so busy workers are still chosen.

  • c
    A longer default cooldown on the group's scale-in scaling policy

    A cooldown only spaces out scaling activities. When scale-in happens, it can still pick a busy worker.

  • d
    Instance scale-in protection that each worker sets and clears per job Correct

    Correct. A protected instance is never chosen for scale-in, so a worker that sets protection when it picks up a job and clears it when idle leaves scale-in to remove only idle workers.

The concept

Instance scale-in protection removes an instance from the candidates for scale-in termination; lifecycle hooks only delay the termination of an instance that has already been selected.

Why that’s the answer

The requirement is about selection: remove idle workers rather than wait for busy ones. Scale-in protection does that — protected instances are skipped. A termination lifecycle hook is the tempting near-twin, but it acts after selection, so a busy worker is still chosen and scale-in waits on it. Termination policies and cooldowns do not know which instances are busy.

How to reason it out
  1. Separate the two problems: which instance is chosen versus how it is shut down.
  2. Have the worker call SetInstanceProtection with protection on when it starts a job.
  3. Clear protection when the job finishes so the idle instance becomes a scale-in candidate.
  4. Keep a lifecycle hook only for clean-up tasks such as shipping logs before termination.

Exam tip: Scale-in protection decides who can be removed; a termination lifecycle hook only delays removal of who was chosen.

Auto Scaling and Elasticity: EC2, Caching, and Database Scaling — the lesson that teaches this.

Question 6Reliability and Business Continuity

A containerized API runs as an Amazon ECS service on Fargate behind an Application Load Balancer. During traffic spikes, response times climb because the service's fixed count of six tasks is insufficient. Which configuration scales the API automatically with the LEAST operational overhead?

Choose one.

  • a
    EC2 Auto Scaling with target tracking for the cluster's container instances

    The service runs on Fargate, so there are no container instances to scale; even on EC2, adding instances does not raise the service's task count.

  • b
    Scheduled actions that raise the service's task count in business hours

    Scheduled actions follow the clock. The spikes are traffic-driven, so a fixed schedule either misses them or pays for idle tasks.

  • c
    A larger task definition size with more CPU and memory for each task

    Bigger tasks raise fixed capacity but still do not respond to load — the service stays at six tasks during spikes and pays for the extra size all day.

  • d
    Service Auto Scaling with target tracking on ALBRequestCountPerTarget Correct

    Correct. ECS service auto scaling adjusts the desired task count through Application Auto Scaling; a target tracking policy on requests per target (or average CPU) follows the spikes with AWS-managed alarms.

The concept

ECS service auto scaling adjusts a service's desired task count via Application Auto Scaling; target tracking on service CPU, memory, or ALBRequestCountPerTarget is the standard configuration.

Why that’s the answer

The service needs more tasks when load rises, which is task-count scaling with target tracking — set a target on ECSServiceAverageCPUUtilization or ALBRequestCountPerTarget and the service follows load automatically. Instance scaling is moot on Fargate, scheduled scaling cannot anticipate unpredictable spikes, and bigger tasks are still a fixed fleet.

How to reason it out
  1. Register the ECS service as a scalable target with Application Auto Scaling.
  2. Choose the metric that best tracks load — ALBRequestCountPerTarget when request volume tracks load better than CPU.
  3. Create a target tracking policy with a sensible target and min/max task counts.
  4. Verify tasks scale out during a spike and back in afterward.

Exam tip: Scenario says tasks or service → ECS service auto scaling with target tracking; on Fargate there are no instances to scale.

Auto Scaling and Elasticity: EC2, Caching, and Database Scaling — the lesson that teaches this.

What SOA-C03 domain 2 tests, topic by topic

The official exam guide breaks Reliability and Business Continuity into 3 topics. The question bank follows the same split, so a weak topic shows up as a cluster of misses you can go back and read.

Published SOA-C03 practice questions per topic in Reliability and Business Continuity
TopicWhat it coversQuestions
Implement scalability and elasticityExam guide task 2.1. Configuring and managing scaling mechanisms across compute environments; caching for dynamic scalability (CloudFront, ElastiCache); scaling in managed databases (RDS, DynamoDB).38
Implement highly available and resilient environmentsExam guide task 2.2. Configuring and troubleshooting Elastic Load Balancing and Route 53 health checks; configuring fault-tolerant systems such as Multi-AZ deployments.38
Implement backup and restore strategiesExam guide task 2.3. Automating snapshots and backups for EC2, RDS, EBS, S3, and DynamoDB with services like AWS Backup; restoring databases (e.g. point-in-time restore) to meet RTO, RPO, and cost requirements; versioning for storage services (S3, FSx); disaster-recovery procedures and best practices (backup and restore, pilot light, warm standby, active/active).38
Total114

Revise Reliability and Business Continuity before you drill it

Other SOA-C03 domains

Reliability and Business Continuity: your questions

Reliability and Business Continuity is domain 2 of the SOA-C03 exam guide and carries 22% of the scored content — the heaviest of the 5 domains. On a 65-question paper that works out to roughly 14 questions, though AWS does not publish an exact per-domain count and individual exam forms vary.

Source

The domain weight and topic list on this page come from the official SOA-C03 exam guide.