Design a solution to ensure business continuity

Tamberlin Media runs a Linux transcoding agent on 200 Amazon EC2 instances across 12 workload accounts. The agent process occasionally hangs, and today an on-call engineer restarts it after a customer complaint, typically 45 minutes later. The company wants the agent restarted automatically within five minutes of a hang, with an audit trail of every remediation, and wants the operations team to see agent health for all 12 accounts from one monitoring account without signing in to each account.

Which solution meets these requirements with the LEAST custom code?

1. AAdd a cron job on each instance that checks the process every minute and restarts it, and forward the cron output to a central Amazon S3 bucket.A per-instance script is custom code on 200 hosts, gives no central health view, and fails silently if the instance itself is impaired.
2. BPublish a procstat metric for the agent with the CloudWatch agent, create a CloudWatch alarm per instance, route alarm state changes through an Amazon EventBridge rule to an AWS Systems Manager Automation runbook that restarts the service, and set up CloudWatch cross-account observability to the monitoring account.The CloudWatch agent's procstat plugin reports process health, alarms detect a hang quickly, and EventBridge can start an Automation runbook whose executions are logged for audit. Cross-account observability links the 12 source accounts to a central monitoring account for unified dashboards.
3. CConfigure an EC2 status check alarm with the recover action on each instance.The recover action responds to system status check failures caused by underlying hardware. A hung application process does not fail the status checks.
4. DEnable AWS Health organizational view and subscribe the on-call team to AWS Health events for the EC2 instances.AWS Health reports AWS-side service events and scheduled changes, not application process failures inside an instance, and it does not perform remediation.

Halvorsen Clinical stores 300 TB of trial documents in an Amazon S3 bucket named trials-primary in us-east-2, encrypted with SSE-KMS. A new contract requires that every new object be present in a bucket in us-west-2 within 15 minutes, backed by a service level agreement, with alerts when replication falls behind. The 300 TB of objects already in the bucket must also be copied to us-west-2.

Which solution meets these requirements?

1. AConfigure S3 Cross-Region Replication without Replication Time Control and rely on the default replication behavior to copy both new and existing objects.Standard replication has no time-based SLA, and a new replication rule does not copy objects that existed before it was created.
2. BConfigure S3 Cross-Region Replication with S3 Replication Time Control and replication metrics enabled, grant the replication role use of the KMS keys in both Regions, and run an S3 Batch Replication job for the existing objects.S3 RTC is designed to replicate 99.99 percent of objects within 15 minutes, is backed by an SLA, and publishes replication metrics and events for alerting. Live replication covers only new objects, so S3 Batch Replication copies the existing ones.
3. CSchedule an AWS DataSync task every 10 minutes from trials-primary to a bucket in us-west-2.DataSync can copy objects but provides no 15-minute replication SLA, and scheduled scans of 300 TB add cost and delay compared with event-driven replication.
4. DEnable S3 Transfer Acceleration on trials-primary and configure an S3 Event Notification to an AWS Lambda function that copies each new object to us-west-2.Transfer Acceleration speeds client uploads over long distances; it does not replicate data. A custom Lambda copier has no SLA and must handle retries and large objects itself.

After a ransomware incident at a peer company, the risk committee at Grenfell Mutual requires that backup copies in the vault central-vault in the account backup-prod be retained for at least 7 years. No user, including the account root user and AWS Backup administrators, may delete the recovery points early or shorten the retention period. The security team wants three days to test the configuration and correct mistakes before it becomes permanent.

Which configuration meets these requirements?

1. AApply AWS Backup Vault Lock in governance mode to central-vault with a minimum retention of 7 years.Governance mode can be removed or changed by users with sufficient IAM permissions, so it does not protect against a compromised administrator or root user.
2. BAttach a service control policy to backup-prod that denies backup:DeleteRecoveryPoint and backup:UpdateRecoveryPointLifecycle.An SCP can be detached or changed by administrators in the management account, and it does not govern lifecycle settings in backup plans, so retention could still be undermined.
3. CAttach a resource-based access policy to central-vault that denies backup:DeleteRecoveryPoint to all principals.A vault access policy helps, but a user with permission to change the policy, or the root user, could edit or remove it, so it does not meet the requirement that no one can bypass it.
4. DApply AWS Backup Vault Lock in compliance mode to central-vault with a minimum retention of 7 years and a grace time of 3 days.In compliance mode, after the grace time (cooling-off period) expires, the lock is immutable: no one, including the root user or AWS, can delete the lock, delete recovery points before their retention ends, or shorten retention. The grace time allows testing and removal before it locks.

Rivenmoor Travel runs its booking web tier on Amazon EC2 instances in an Auto Scaling group spread across three Availability Zones behind an Application Load Balancer in us-east-1. During a recent partial Availability Zone impairment, instances in one zone returned slow responses while still passing load balancer health checks, and users saw timeouts for 40 minutes. The company wants traffic to move away from an impaired Availability Zone automatically when AWS detects an issue in that zone, and to be able to shift traffic away from a zone on demand during its own incidents.

Which solution meets these requirements?

1. AEnable zonal shift on the Application Load Balancer, enable zonal autoshift for it in Amazon Application Recovery Controller (ARC) with the required practice run configuration, keep enough capacity in the Auto Scaling group to serve full load from two zones, and use zonal shift to move traffic manually when needed.Application Load Balancers, with cross-zone load balancing enabled or disabled, are supported ARC resources. Zonal shift moves traffic away from one Availability Zone on demand, and zonal autoshift lets AWS start the shift automatically when it detects an impairment in that zone; ARC requires a practice run configuration for autoshift so the shift is exercised regularly. Pre-provisioning capacity so two zones can carry the load makes the shift safe.
2. BShorten the target group health check interval and lower the unhealthy threshold so that slow instances are marked unhealthy sooner.More aggressive health checks can cause flapping and still depend on instances failing the check, which did not happen during the gray failure. They also provide no way to evacuate a zone on demand.
3. CDeploy a second ALB in us-west-2 and configure Amazon Route 53 failover routing between the two Regions.Regional failover is a much larger and costlier response to a single-zone problem, and Route 53 health checks would also have passed during this gray failure.
4. DTurn on cross-zone load balancing for the target group so that every load balancer node distributes traffic evenly across all instances.Cross-zone load balancing only controls whether load balancer nodes spread requests across targets in all zones. It is not a mechanism for removing a zone, so traffic would still reach the impaired instances.