Practice Test #9

Instance store is not a direct target for restoring EBS snapshots and would require extra copying or rebuild work for 50 TB. Trap: Instance store sounds fast, but the issue is cloning EBS snapshot data quickly and safely into staging.

EBS Multi-Attach shares the same production volumes and violates the requirement for complete isolation from production. Trap: It sounds quick because no data is copied, but sharing production storage creates testing impact and corruption risk.

New EBS volumes from snapshots are isolated, but without Fast Snapshot Restore they are lazy-loaded and may need lengthy initialization before delivering high IOPS. Trap: This is close, but standard snapshot restore does not minimize readiness time for a 50 TB, high-IOPS workload.

EC2 with Docker and Auto Scaling can add instances, but the team still owns Docker hosts, patching, deployment mechanics, and container scheduling. Trap: It sounds elastic because of Auto Scaling Groups, but it does not remove infrastructure management.

ECS with the EC2 launch type gives managed orchestration, but the EC2 cluster still needs capacity planning, patching, and scaling coordination. Trap: Cluster auto scaling helps, but it is still more operationally heavy than Fargate.

Manually configuring orchestration on ECS-optimized EC2 instances leaves the most hands-on work and conflicts with the request to focus on application logic. Trap: The ECS-optimized AMI sounds purpose-built, but manual orchestration is the opposite of the low-ops requirement.

Auto Scaling can add EC2 capacity and the notification half is valid, but it keeps the custom polling fleet, scaling policies, patching, and duplicate-work coordination. Trap: High CPU points toward scaling, but the question prioritizes least operational overhead.

EventBridge routes events but does not inherently pull records from each EHR API into S3. A producer or custom integration would still be required. Trap: EventBridge sounds serverless, but it is not a replacement for managed API data ingestion here.

ECS can scale the application, but the company would still operate containers, task definitions, scaling policies, monitoring, and the same polling workflow. Trap: Containers improve deployment, but they do not remove the ingestion management burden.

S3 Standard is durable and multi-AZ, but it does not reduce storage cost for documents that sit dormant for months. Trap: It is safe and simple, but the scenario explicitly asks to minimize cost for mixed access.

S3 Standard-IA is multi-AZ, but frequent or unpredictable audit access can create retrieval charges and erase the savings. Trap: It looks cheaper than Standard, but it assumes access is reliably infrequent.

S3 One Zone-IA stores data in one Availability Zone, which fails the requirement to survive the complete failure of any single AZ. Trap: Its lower price is attractive, but it trades away the resilience the scenario requires.

Lambda is reasonable for processing, but DynamoDB is not appropriate for storing video files because it is designed for items and metadata, not large media objects. Trap: It pairs serverless compute with a scalable database, but the storage target is wrong for videos.

Kinesis Data Firehose is a managed delivery service for streaming records; it does not run arbitrary thumbnail generation or act as a primary metadata database. Trap: The word streaming may seem related to video, but Firehose is not a video processing platform.

A fixed fleet of five EC2 instances with EBS volumes does not automatically match 50-to-2000-user swings and adds server and storage operations. Trap: More EC2 capacity sounds like scaling, but it is static and operationally heavier.

CloudFront is a CDN for HTTP/HTTPS delivery and caching; it is not the right service for accelerating arbitrary RTMP TCP ingestion into a broadcast system. Trap: It is an edge service, but this is TCP acceleration rather than web content distribution.

Client VPN provides private access to a VPC, but it does not provide global anycast ingress or AWS backbone acceleration for public RTMP streams. Trap: VPN sounds like a network fix, but it can add overhead and is not a media acceleration service.

EC2 instances with Elastic IPs still depend on standard internet routing to us-east-1 and do not solve global congestion or latency. Trap: Elastic IPs provide stable addresses, not accelerated global routing.

Manually reviewing each bucket in the S3 console does not scale to 50+ buckets and multiple Regions. It introduces ongoing operational burden, inconsistency, and a high risk of missing trends across accounts. Trap: Monitor individual bucket metrics through the S3 console dashboard and manually review access... sounds related, but the scenario is really testing managed S3-wide cost optimization with minimal operations.

CloudWatch detailed monitoring for S3 does not inherently provide the kind of cross-bucket, cost-optimization analytics needed to identify rarely accessed content at scale. Building custom queries and dashboards in QuickSight adds significant operational overhead (data collection, modeling, permissions, and maintenance). Trap: Enable CloudWatch detailed monitoring for S3 bucket metrics and create custom queries using Amazon... sounds related, but the scenario is really testing managed S3-wide cost optimization with minimal operations.

), which could be analyzed to infer access frequency, but enabling data events for all buckets is costly and generates large volumes of logs. Processing those logs with EMR adds substantial operational complexity (cluster management, ETL, ongoing tuning). Trap: Implement AWS CloudTrail data events for all S3 buckets and process the logs using Amazon EMR to... sounds related, but the scenario is really testing managed S3-wide cost optimization with minimal operations.

FSx for Lustre is high performance, but the option names a Multi-AZ deployment model that is not the standard Lustre answer here and does not emphasize elastic NFS hybrid access. Trap: Media editing can point toward Lustre, but the question stresses elastic shared Linux file storage and hybrid mounts.

EBS Multi-Attach is shared block storage with strict limits and requires clustered file-system handling; it is not a managed shared file system for many workstations. Trap: Multi-Attach sounds shared, but block access is not the same as NFS file access.

A single EFS mount target creates an AZ dependency and does not meet the high availability requirement. Trap: Access points help organize permissions, but they do not replace multi-AZ mount targets.

This option is incorrect because resource state such as stopped or terminated is not the primary criterion for backup coverage. Terminated resources generally cannot be backed up, and stopped resources are not discovered or protected simply by placing them in centralized backup vaults. Trap: It may look plausible because it mentions AWS Config, but the scenario is about automated backup enrollment by tags.

This option is incorrect because quarterly manual audits create significant operational overhead and are highly prone to human error. In a 15-account healthcare environment with HIPAA requirements, relying on department heads to identify resources would likely miss newly created or changed resources between audit cycles. Trap: It may look plausible because it mentions a related AWS feature, but the scenario is about automated backup enrollment by tags.

This option is incorrect because Amazon Inspector is a security assessment and vulnerability management service, not a backup governance or resource enrollment tool. Inspector can identify software vulnerabilities and exposure issues, but it does not assign resources to AWS Backup plans or enforce backup coverage. Trap: It may look plausible because it mentions Amazon Inspector, but the scenario is about automated backup enrollment by tags.