Design Resilient Architectures

54. A payment service must not lose messages when its consumer fails, and failed messages must be reprocessed later without blocking new work. Which solution meets these requirements?

1. A Send messages to an Amazon SQS queue and increase the visibility timeout so failed messages are retried later.
2. B Write messages to an Amazon Kinesis Data Stream with a one-hour retention period.
3. C Send messages to an Amazon SQS queue and configure a redrive policy that moves messages to a dead-letter queue after a maximum receive count.
4. D Publish messages to an Amazon SNS topic and subscribe the consumer to the topic.

55. Messages in an Amazon SQS queue are occasionally processed twice because a consumer sometimes takes longer than expected to finish. Which combination of steps meets these requirements? (Select TWO.)

1. A Reduce the queue's message retention period so stale messages expire.
2. B Increase the queue's visibility timeout to exceed the longest expected processing time.
3. C Decrease the queue's visibility timeout so failed messages are retried sooner.
4. D Modify the consumer so that processing the same message twice produces the same result.
5. E Increase the maximum receive count in the queue's redrive policy.

56. An application must continue accepting orders when a downstream fulfilment service becomes unavailable, and must process the accepted orders once that service recovers. Which solution meets these requirements?

1. A Call the fulfilment service synchronously and configure a short client timeout with retries.
2. B Call the fulfilment service synchronously and retry failed calls until they succeed.
3. C Write order events to an Amazon SQS queue that the fulfilment service consumes independently.
4. D Run the order service and the fulfilment service in the same Auto Scaling group.

57. An AWS Lambda function invoked asynchronously by Amazon S3 events must not discard events that fail after all retry attempts. Which solution meets these requirements?

1. A Increase the function's memory allocation so invocations are less likely to fail.
2. B Configure an on-failure destination on the function's asynchronous invocation settings.
3. C Configure reserved concurrency on the function so invocations are not throttled.
4. D Increase the function timeout so slow invocations complete before failing.

58. An AWS Step Functions workflow calls an external API that is intermittently unavailable. Transient failures must not fail the whole execution. Which solution meets these requirements?

1. A Wrap the API call in a Parallel state that invokes it twice.
2. B Increase the state machine's overall execution timeout.
3. C Configure an Amazon EventBridge rule that restarts the execution when it fails.
4. D Add a Retry policy with exponential backoff to that state and a Catch clause for terminal failures.

59. Orders are processed through an Amazon SQS queue. During a prolonged downstream outage, messages older than the retention period were lost. Which combination of steps meets these requirements? (Select TWO.)

1. A Create an Amazon CloudWatch alarm on the age of the oldest message in the queue.
2. B Increase the queue's message retention period to cover the longest tolerable outage.
3. C Decrease the queue's visibility timeout so messages are redelivered sooner.
4. D Convert the queue to a FIFO queue with content-based deduplication.
5. E Increase the number of consumer instances during the outage.

60. A queue-backed worker fleet must finish processing the message in flight before Auto Scaling terminates an instance during scale-in. Which solution meets these requirements?

1. A Increase the Auto Scaling group minimum size so scale-in occurs less often.
2. B Reduce the queue's visibility timeout so an interrupted message reappears sooner.
3. C Enable termination protection on every instance in the Auto Scaling group.
4. D Add an Auto Scaling lifecycle hook that holds the instance in a terminating wait state until the worker signals completion.

61. An application runs on a single Amazon EC2 instance that writes to one Amazon EBS volume. The business requires the application to survive the loss of an Availability Zone with a recovery time objective of minutes. Which solution meets these requirements?

1. A Increase the frequency of EBS snapshots and store them in the same Availability Zone.
2. B Launch a larger instance type in the same Availability Zone and enable termination protection.
3. C Deploy the application in an Auto Scaling group in a single Availability Zone behind an Application Load Balancer and keep state on the EBS volume.
4. D Deploy the application in an Auto Scaling group across multiple Availability Zones behind an Application Load Balancer and move shared state to Amazon EFS.

62. An Amazon RDS for PostgreSQL database must fail over automatically to another Availability Zone with the least possible data loss. Which solution meets these requirements?

1. A Change the instance class and attach Provisioned IOPS storage to reduce the chance of failure.
2. B Create a read replica in a second Availability Zone and promote it when the primary fails.
3. C Convert the instance to a Multi-AZ deployment so a synchronous standby is maintained in a second Availability Zone.
4. D Increase the automated backup retention period to 35 days and restore to a point in time when the primary fails.

63. Multiple Amazon EC2 instances across several Availability Zones must mount the same file system with read and write access at the same time. Which solution meets these requirements?

1. A Create an Amazon EFS file system and mount it from instances in each Availability Zone.
2. B Create an Amazon EBS gp3 volume and attach it to instances in each Availability Zone.
3. C Create an Amazon EBS volume with Multi-Attach enabled and attach it to instances in each Availability Zone.
4. D Store the data in Amazon S3 and mount the bucket as a drive on each instance.

64. A workload must survive the loss of an entire Region with a recovery point objective near zero and a recovery time objective of a few minutes, at the lowest cost that still meets those targets. Which solution meets these requirements?

1. A Replicate data to a second Region and provision the application tier only when failover is required.
2. B Copy database snapshots to a second Region on a nightly schedule and restore them when failover is required.
3. C Run a scaled-down copy of the environment in a second Region with continuous data replication, and scale it up during failover.
4. D Run a full-capacity copy of the environment in a second Region and serve traffic from both Regions.

65. A stateless web tier behind an Application Load Balancer must lose no serving capacity when one Availability Zone fails. Peak load requires 8 instances. Which solution meets these requirements at the lowest instance count?

1. A Run 8 instances distributed evenly across two Availability Zones.
2. B Run 12 instances distributed evenly across three Availability Zones.
3. C Run 16 instances distributed evenly across two Availability Zones.
4. D Run 8 instances distributed evenly across three Availability Zones.

66. An application writes to an Amazon Aurora cluster. The business requires that the loss of a Region be survivable with a recovery point objective under one minute. Which solution meets these requirements?

1. A Enable Aurora Backtrack and rewind the cluster during failover.
2. B Configure a Multi-AZ Aurora cluster with three Aurora Replicas in the primary Region.
3. C Configure an Aurora global database with a secondary Region and promote the secondary during failover.
4. D Copy automated snapshots to a second Region on a nightly schedule.

67. An Auto Scaling group repeatedly launches and terminates instances in a short cycle as load fluctuates around the scaling threshold. Which solution meets these requirements?

1. A Change the load balancer from an Application Load Balancer to a Network Load Balancer.
2. B Lower the scaling threshold so the group scales out sooner.
3. C Add a cooldown period, or use a target tracking policy with an instance warm-up value.
4. D Reduce the group's minimum size so fewer instances are maintained.

68. A critical batch job runs nightly on one Amazon EC2 instance. If the instance fails partway through, the job must resume rather than restart from the beginning. Which solution meets these requirements?

1. A Enable termination protection on the instance so it cannot be terminated during the run.
2. B Modify the job to write progress checkpoints to Amazon S3 and resume from the last checkpoint on a replacement instance.
3. C Launch a larger instance type so the job completes before a failure is likely.
4. D Create an EBS snapshot before the job starts and restore it after a failure.

69. A static website hosted in Amazon S3 must continue serving read traffic if the primary Region becomes unavailable. Which combination of steps meets these requirements? (Select TWO.)

1. A Copy EBS snapshots of the web servers to the second Region.
2. B Configure S3 Cross-Region Replication to a bucket in a second Region.
3. C Create Amazon Route 53 failover records with a health check on the primary endpoint.
4. D Enable S3 Versioning on the primary bucket.
5. E Create an Application Load Balancer that spans both Regions.

70. Read traffic against an Amazon Aurora cluster varies unpredictably between the load of one replica and twelve. Which solution meets these requirements at the lowest cost?

1. A Provision twelve Aurora Replicas and direct all read traffic to the reader endpoint.
2. B Convert the writer instance to Aurora Serverless v2 and direct all read traffic to it.
3. C Provision one Aurora Replica on a larger instance class and direct all read traffic to it.
4. D Configure Aurora Auto Scaling to add and remove Aurora Replicas against a target metric.

71. An Amazon DynamoDB table must be restorable to any second within the previous 35 days after an accidental bulk delete. Which solution meets these requirements?

1. A Enable DynamoDB Streams and replay the change records after a deletion.
2. B Enable point-in-time recovery on the table.
3. C Configure the table as a global table with a replica in a second Region.
4. D Create on-demand backups of the table on a weekly schedule.

72. An application must survive the failure of an entire Region with a recovery time objective under five minutes and a recovery point objective near zero. Cost is not the primary constraint. Which solution meets these requirements?

1. A Run a scaled-down copy in a second Region and scale it up during failover.
2. B Replicate data to a second Region and provision the application tier during failover.
3. C Restore the application from cross-Region snapshots during failover.
4. D Run the application at full capacity in two Regions and distribute traffic to both.

73. An Auto Scaling group replaces instances that fail EC2 status checks, but leaves instances running when the application process crashes while the operating system remains reachable. Which solution meets these requirements?

1. A Increase the Auto Scaling group desired capacity so healthy instances absorb the traffic.
2. B Reduce the Auto Scaling group health check grace period so failures are detected sooner.
3. C Change the Auto Scaling group health check type to ELB so load balancer target health is used.
4. D Reduce the load balancer health check interval so failures are detected sooner.

74. On-premises users require low-latency access to a shared file set whose authoritative copy must reside in Amazon S3 and survive an on-premises outage. Which solution meets these requirements?

1. A Mount an Amazon EFS file system from the on-premises network over the internet.
2. B Deploy an Amazon EC2 instance running a file server with Amazon EBS storage and copy files to Amazon S3 nightly.
3. C Deploy AWS Storage Gateway in file gateway mode with a local cache backed by Amazon S3.
4. D Deploy AWS DataSync to copy the files to Amazon S3 on an hourly schedule.

75. An application deployed in two Regions must send users to the Region with the lowest latency and must stop sending users to a Region that fails its health check. Which solution meets these requirements?

1. A Create Amazon Route 53 latency records for both endpoints without associating health checks.
2. B Create Amazon Route 53 simple records that return both endpoint addresses.
3. C Create Amazon Route 53 latency records for both endpoints and associate a health check with each record.
4. D Create Amazon Route 53 weighted records that split traffic evenly between the endpoints.

76. An Amazon EFS file system holds data where recent files are accessed constantly and older files are rarely read. All data must remain immediately available. Which solution meets these requirements at the lowest cost?

1. A Delete files that have not been accessed for 90 days and restore them from backup on request.
2. B Create the file system with the One Zone storage class to reduce the storage price.
3. C Reduce the file system's provisioned throughput to lower the monthly charge.
4. D Configure an EFS lifecycle policy that transitions files to the Infrequent Access storage class after a period without access.

77. A stateless application must lose no capacity when one of two Availability Zones fails. Peak load requires 10 instances. Which solution meets these requirements?

1. A Run 15 instances distributed evenly across the two Availability Zones.
2. B Run 10 instances distributed evenly across the two Availability Zones.
3. C Run 20 instances distributed evenly across the two Availability Zones.
4. D Run 10 instances in one Availability Zone and keep an Amazon Machine Image ready in the other.

78. A critical object in Amazon S3 must not be permanently deletable by a compromised credential, while routine updates to the object must continue. Which combination of steps meets these requirements? (Select TWO.)

1. A Add a bucket policy that denies s3:PutObject to all principals.
2. B Enable S3 Versioning on the bucket so a delete places a marker and prior versions are retained.
3. C Enable S3 Object Lock in compliance mode with a long retention period.
4. D Enable MFA Delete on the bucket so permanent version deletion requires multi-factor authentication.
5. E Configure S3 Cross-Region Replication to a bucket in a second Region.

79. An application's web tier must not fail when its processing tier is unavailable. Which solution meets these requirements?

1. A Call the processing tier synchronously with a short timeout.
2. B Place an Amazon SQS queue between the tiers so the web tier writes messages the processing tier consumes independently.
3. C Run both tiers in the same Auto Scaling group.
4. D Increase the processing tier's instance count.

80. A workload must scale its compute capacity based on the number of messages waiting in a queue. Which solution meets these requirements?

1. A Create a scheduled scaling action at fixed times.
2. B Create a target tracking scaling policy on a custom metric of backlog per instance derived from the queue depth.
3. C Create a target tracking scaling policy on average CPU utilization.
4. D Set a fixed instance count sized for the largest observed backlog.

81. Several microservices must react to the same business event without the publisher knowing about them. Which solution meets these requirements?

1. A Send the event to a single SQS queue all services poll.
2. B Publish the event to Amazon EventBridge and have each service create a rule matching it.
3. C Write the event to a shared database each service polls.
4. D Call each service in turn from the publisher.

82. A workload processes long-running jobs that must survive an instance failure mid-job. Which solution meets these requirements?

1. A Increase the instance size so failures are less likely.
2. B Store job state externally and use a queue with a visibility timeout so an unfinished job returns for reprocessing.
3. C Store job state on the instance's local disk.
4. D Process jobs synchronously from the web tier.

83. An application must run a series of steps with branching and error handling, and each step's status must be visible. Which solution meets these requirements?

1. A Chain Lambda functions so each invokes the next.
2. B Run all steps inside a single Lambda function.
3. C Schedule each step with a separate EventBridge rule.
4. D Orchestrate the steps as an AWS Step Functions state machine.

84. A relational database must remain available if its Availability Zone fails, with automatic failover. Which solution meets these requirements?

1. A Deploy the database on a larger instance class.
2. B Deploy the database with a read replica in a second Availability Zone.
3. C Deploy the database as a Multi-AZ deployment with a standby in a second Availability Zone.
4. D Take automated snapshots every hour.

85. A stateless web tier must remain available during an Availability Zone failure. Which solution meets these requirements?

1. A Run a larger instance in one Availability Zone.
2. B Run the instances in an Auto Scaling group spanning multiple Availability Zones behind a load balancer.
3. C Run the instances in one Availability Zone with hourly snapshots.
4. D Run the instances in one Availability Zone with a standby AMI.

86. An application's file storage must be accessible from instances in several Availability Zones simultaneously. Which solution meets these requirements?

1. A Amazon EBS, which attaches to one instance in one zone.
2. B Amazon S3 mounted as a local file system.
3. C Amazon EFS, which provides a shared file system across Availability Zones.
4. D Instance store volumes on each instance.

87. A solution must direct users to a secondary Region if the primary Region's application becomes unhealthy. Which solution meets these requirements?

1. A Create Route 53 simple records returning both endpoints.
2. B Reduce the DNS record's time to live.
3. C Create Route 53 weighted records split between the Regions.
4. D Create Route 53 failover records with a health check against the primary Region's application endpoint.

88. An application must tolerate the loss of an Availability Zone without losing queued work. Which characteristic of Amazon SQS supports this?

1. A Messages are stored redundantly across multiple Availability Zones within the Region.
2. B Messages are held in the producer's memory until consumed.
3. C Messages are stored on a single instance with a snapshot.
4. D Messages are replicated to a second Region automatically.

89. An application's components must communicate without either knowing the other's location. Which solution meets these requirements?

1. A Deploy both components on the same instance.
2. B Have each component call the other's IP address directly.
3. C Publish messages to a queue or topic so producers and consumers are unaware of each other.
4. D Configure DNS records for each component.

90. A workload must process messages exactly once where duplicates would cause incorrect results. Which solution meets these requirements?

1. A Use a standard queue and accept occasional duplicates.
2. B Use a standard queue with a longer visibility timeout.
3. C Use an SQS FIFO queue with content-based or explicit deduplication, and make the consumer idempotent.
4. D Use an SNS topic with several subscribers.

91. An application must invoke a long-running process without the caller waiting for it to complete. Which solution meets these requirements?

1. A Increase the caller's timeout so the process completes.
2. B Return an error if the process takes too long.
3. C Accept the request, place the work on a queue, and return an identifier the caller uses to check the result.
4. D Run the process synchronously on a larger instance.

92. A microservice must be able to scale independently of the services that call it. Which characteristic is required?

1. A It must run on the same instances as its callers.
2. B It must share a database with its callers.
3. C It must be stateless, or hold state externally, so instances can be added and removed freely.
4. D It must maintain session affinity with each caller.

93. An application must react to changes in a DynamoDB table without polling it. Which solution meets these requirements?

1. A Query the table on a schedule for changed items.
2. B Add a last-modified attribute and scan for recent changes.
3. C Increase the table's provisioned read capacity.
4. D Enable DynamoDB Streams and process the change records with a Lambda function.

94. A workload's components must be able to fail independently without the failure cascading. Which pattern applies?

1. A A single process containing every component.
2. B Synchronous calls with retries between components.
3. C A shared thread pool sized for the busiest component.
4. D Bulkhead isolation, giving each component its own resource pool.

95. An application must distribute work across consumers so each message is processed once. Which solution meets these requirements?

1. A Write the work to Amazon S3 for consumers to list.
2. B Use EventBridge with several rules matching the same event.
3. C Use an SQS queue that several consumers poll, since each message is delivered to one consumer.
4. D Use an SNS topic with several subscribers.

96. An application must buffer requests during a traffic spike so the backend processes them at a sustainable rate. Which pattern applies?

1. A Scaling the backend to the spike's peak.
2. B Queue-based load levelling, with the queue absorbing the spike.
3. C Rejecting requests that exceed the backend's capacity.
4. D Increasing the backend's request timeout.

97. A workflow's steps must run in order with error handling visible per step. Which solution meets these requirements?

1. A Chain Lambda functions so each invokes the next.
2. B Run all steps in one Lambda function.
3. C Define the workflow as a Step Functions state machine with Retry and Catch on each state.
4. D Trigger each step from a separate EventBridge schedule.

98. An Aurora cluster must continue serving reads if its writer instance fails. Which configuration is required?

1. A Enable Performance Insights on the cluster.
2. B Enable automated backups on the cluster.
3. C Add at least one reader instance in another Availability Zone so it can be promoted.
4. D Increase the writer instance's size.

99. An application must continue accepting requests while a downstream component is being deployed. Which solution meets these requirements?

1. A Call the downstream component synchronously with retries.
2. B Increase the downstream component's instance count before deploying.
3. C Deploy the downstream component during a maintenance window.
4. D Place a queue between the components so requests are buffered during the deployment.