A reports an inadequate or failed acceptance condition: Storage-class inventory shows long-lived objects never leave S3 Standard. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that storage-class inventory matches aging policy and end-to-end retrieval cost stays within target. The underlying design requirement is lifecycle transitions reduce storage charges when access frequency and retrieval penalties justify the tier.

C reports an inadequate or failed acceptance condition: Deep Archive early-deletion and retrieval charges exceed the projected class savings. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that storage-class inventory matches aging policy and end-to-end retrieval cost stays within target. The underlying design requirement is lifecycle transitions reduce storage charges when access frequency and retrieval penalties justify the tier.

D reports an inadequate or failed acceptance condition: Lifecycle configuration expires an object before the required retention date. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that storage-class inventory matches aging policy and end-to-end retrieval cost stays within target. The underlying design requirement is lifecycle transitions reduce storage charges when access frequency and retrieval penalties justify the tier.

A would buy Compute Savings Plans expecting them to offset cross-AZ network transfer line items. This is not a defensible answer to the stated constraint (A traffic-heavy application copies 3 TB daily between AZs without a resilience reason): aZ data transfer cost can dominate network spend; placement optimization cannot discard required fault tolerance. The required design is identify cross-AZ transfer charges in CUR and reduce unnecessary chatty inter-AZ hops; acceptance depends on evidence that itemized transfer spend falls while mandatory multi-AZ failure tolerance remains tested.

B would route even more chatty workload exchanges over an inter-Region link instead of reducing unnecessary hops. This is not a defensible answer to the stated constraint (A traffic-heavy application copies 3 TB daily between AZs without a resilience reason): aZ data transfer cost can dominate network spend; placement optimization cannot discard required fault tolerance. The required design is identify cross-AZ transfer charges in CUR and reduce unnecessary chatty inter-AZ hops; acceptance depends on evidence that itemized transfer spend falls while mandatory multi-AZ failure tolerance remains tested.

D would move all app components to one AZ and abandon necessary failure tolerance. This is not a defensible answer to the stated constraint (A traffic-heavy application copies 3 TB daily between AZs without a resilience reason): aZ data transfer cost can dominate network spend; placement optimization cannot discard required fault tolerance. The required design is identify cross-AZ transfer charges in CUR and reduce unnecessary chatty inter-AZ hops; acceptance depends on evidence that itemized transfer spend falls while mandatory multi-AZ failure tolerance remains tested.

A reports an inadequate or failed acceptance condition: A Savings Plan bill reflects compute discounts while network transfer costs remain unchanged. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that itemized transfer spend falls while mandatory multi-AZ failure tolerance remains tested. The underlying design requirement is aZ data transfer cost can dominate network spend; placement optimization cannot discard required fault tolerance.

B reports an inadequate or failed acceptance condition: Inter-Region hops were added and the billed transfer rate rose rather than fell. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that itemized transfer spend falls while mandatory multi-AZ failure tolerance remains tested. The underlying design requirement is aZ data transfer cost can dominate network spend; placement optimization cannot discard required fault tolerance.

C reports an inadequate or failed acceptance condition: Cross-AZ transfer bytes decline to zero only because all replicas share one failure domain. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that itemized transfer spend falls while mandatory multi-AZ failure tolerance remains tested. The underlying design requirement is aZ data transfer cost can dominate network spend; placement optimization cannot discard required fault tolerance.

B would stop compute workers but leave associated public IP allocations and unused storage billing unattended. This is not a defensible answer to the stated constraint (A service leaves expensive storage and public IPs allocated long after development jobs exit): cleanup of unused billed resources saves money without changing active architecture when retention is respected. The required design is use owner-tagged inventory and automated cleanup of unused resources with preservation exceptions; acceptance depends on evidence that nonproduction orphan costs disappear while retained customer data remains recoverable.

C would delete every unattached volume with no owner or data-retention exception checks. This is not a defensible answer to the stated constraint (A service leaves expensive storage and public IPs allocated long after development jobs exit): cleanup of unused billed resources saves money without changing active architecture when retention is respected. The required design is use owner-tagged inventory and automated cleanup of unused resources with preservation exceptions; acceptance depends on evidence that nonproduction orphan costs disappear while retained customer data remains recoverable.

D would reserve expensive resources indefinitely because no tag identifies their previous owner. This is not a defensible answer to the stated constraint (A service leaves expensive storage and public IPs allocated long after development jobs exit): cleanup of unused billed resources saves money without changing active architecture when retention is respected. The required design is use owner-tagged inventory and automated cleanup of unused resources with preservation exceptions; acceptance depends on evidence that nonproduction orphan costs disappear while retained customer data remains recoverable.

A reports an inadequate or failed acceptance condition: All compute jobs stopped but unused storage remains attached to no active workloads. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that nonproduction orphan costs disappear while retained customer data remains recoverable. The underlying design requirement is cleanup of unused billed resources saves money without changing active architecture when retention is respected.

C reports an inadequate or failed acceptance condition: Orphan counts fall only after a protected recovery volume was deleted. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that nonproduction orphan costs disappear while retained customer data remains recoverable. The underlying design requirement is cleanup of unused billed resources saves money without changing active architecture when retention is respected.

D reports an inadequate or failed acceptance condition: An idle public address is still allocated and billed because it lacks an owner tag. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that nonproduction orphan costs disappear while retained customer data remains recoverable. The underlying design requirement is cleanup of unused billed resources saves money without changing active architecture when retention is respected.

A would compare only EC2 instance hours and ignore storage, replication, transfer, and failover costs. This is not a defensible answer to the stated constraint (A company needs to forecast the cost difference between always-on regional replicas and warm standby): a cost model must include replication and transfer, not only the secondary instance hourly price. The required design is model compute, storage, transfer, and recovery costs with AWS Pricing Calculator and CUR validation; acceptance depends on evidence that a documented forecast compares steady and failover costs under agreed usage assumptions.

B would use last month's billed spend as a complete forecast for both untested deployment patterns. This is not a defensible answer to the stated constraint (A company needs to forecast the cost difference between always-on regional replicas and warm standby): a cost model must include replication and transfer, not only the secondary instance hourly price. The required design is model compute, storage, transfer, and recovery costs with AWS Pricing Calculator and CUR validation; acceptance depends on evidence that a documented forecast compares steady and failover costs under agreed usage assumptions.

D would model the warm standby as continuously active at full primary-region capacity without agreed usage assumptions. This is not a defensible answer to the stated constraint (A company needs to forecast the cost difference between always-on regional replicas and warm standby): a cost model must include replication and transfer, not only the secondary instance hourly price. The required design is model compute, storage, transfer, and recovery costs with AWS Pricing Calculator and CUR validation; acceptance depends on evidence that a documented forecast compares steady and failover costs under agreed usage assumptions.

A reports an inadequate or failed acceptance condition: The estimate ignores cross-Region replication storage and internet transfer for the active-active design. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that a documented forecast compares steady and failover costs under agreed usage assumptions. The underlying design requirement is a cost model must include replication and transfer, not only the secondary instance hourly price.

B reports an inadequate or failed acceptance condition: The previous monthly bill provides no failover-state pricing under projected demand. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that a documented forecast compares steady and failover costs under agreed usage assumptions. The underlying design requirement is a cost model must include replication and transfer, not only the secondary instance hourly price.

C reports an inadequate or failed acceptance condition: The warm standby estimate assumes full always-on production traffic and overstates its real idle baseline. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that a documented forecast compares steady and failover costs under agreed usage assumptions. The underlying design requirement is a cost model must include replication and transfer, not only the secondary instance hourly price.

B would use CPU utilization alerts alone and assume they predict streaming API and data-ingestion charges. This is not a defensible answer to the stated constraint (A team wants alerts before a variable-cost streaming pipeline exceeds an approved monthly ceiling): budgets and anomaly detection warn about overspend; ordinary CloudWatch host CPU alarms do not. The required design is use AWS Budgets with forecasted thresholds and Anomaly Detection for unexpected spikes; acceptance depends on evidence that finance receives spend and anomaly alerts ahead of the cost policy threshold.

C would review the final monthly bill only after the approved streaming ceiling has been exceeded. This is not a defensible answer to the stated constraint (A team wants alerts before a variable-cost streaming pipeline exceeds an approved monthly ceiling): budgets and anomaly detection warn about overspend; ordinary CloudWatch host CPU alarms do not. The required design is use AWS Budgets with forecasted thresholds and Anomaly Detection for unexpected spikes; acceptance depends on evidence that finance receives spend and anomaly alerts ahead of the cost policy threshold.

D would configure AWS Budgets with no recipient or alert threshold for pipeline spend. This is not a defensible answer to the stated constraint (A team wants alerts before a variable-cost streaming pipeline exceeds an approved monthly ceiling): budgets and anomaly detection warn about overspend; ordinary CloudWatch host CPU alarms do not. The required design is use AWS Budgets with forecasted thresholds and Anomaly Detection for unexpected spikes; acceptance depends on evidence that finance receives spend and anomaly alerts ahead of the cost policy threshold.

A reports an inadequate or failed acceptance condition: A CPU alarm is quiet while streaming ingestion charges double overnight. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that finance receives spend and anomaly alerts ahead of the cost policy threshold. The underlying design requirement is budgets and anomaly detection warn about overspend; ordinary CloudWatch host CPU alarms do not.

C reports an inadequate or failed acceptance condition: Anomaly reports arrive after the monthly billing period has already closed. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that finance receives spend and anomaly alerts ahead of the cost policy threshold. The underlying design requirement is budgets and anomaly detection warn about overspend; ordinary CloudWatch host CPU alarms do not.

D reports an inadequate or failed acceptance condition: A Budget record exists but no recipient receives an alert before the ceiling. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that finance receives spend and anomaly alerts ahead of the cost policy threshold. The underlying design requirement is budgets and anomaly detection warn about overspend; ordinary CloudWatch host CPU alarms do not.

A would attach gateway endpoints only to a separate VPC and expect the workload's route tables to change. This is not a defensible answer to the stated constraint (A microservice's NAT data processing costs exceed compute costs because it calls S3 and DynamoDB from private subnets): gateway endpoints avoid per-byte NAT processing for eligible regional S3 and DynamoDB traffic. The required design is use appropriate S3 and DynamoDB gateway VPC endpoints with route associations; acceptance depends on evidence that nAT processed-byte metrics decline while service calls succeed without internet routes.

B would leave all S3 and DynamoDB private-subnet calls crossing a metered NAT gateway. This is not a defensible answer to the stated constraint (A microservice's NAT data processing costs exceed compute costs because it calls S3 and DynamoDB from private subnets): gateway endpoints avoid per-byte NAT processing for eligible regional S3 and DynamoDB traffic. The required design is use appropriate S3 and DynamoDB gateway VPC endpoints with route associations; acceptance depends on evidence that nAT processed-byte metrics decline while service calls succeed without internet routes.

D would create interface endpoints for unrelated APIs while keeping S3/DynamoDB gateway routes on NAT. This is not a defensible answer to the stated constraint (A microservice's NAT data processing costs exceed compute costs because it calls S3 and DynamoDB from private subnets): gateway endpoints avoid per-byte NAT processing for eligible regional S3 and DynamoDB traffic. The required design is use appropriate S3 and DynamoDB gateway VPC endpoints with route associations; acceptance depends on evidence that nAT processed-byte metrics decline while service calls succeed without internet routes.

A reports an inadequate or failed acceptance condition: Gateway endpoint route tables cover one test subnet but not the private workload subnets. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that nAT processed-byte metrics decline while service calls succeed without internet routes. The underlying design requirement is gateway endpoints avoid per-byte NAT processing for eligible regional S3 and DynamoDB traffic.

B reports an inadequate or failed acceptance condition: The application S3 calls still traverse NAT and its processed-byte bill is unchanged. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that nAT processed-byte metrics decline while service calls succeed without internet routes. The underlying design requirement is gateway endpoints avoid per-byte NAT processing for eligible regional S3 and DynamoDB traffic.

C reports an inadequate or failed acceptance condition: An endpoint for a different service reports Available but S3 and DynamoDB traffic bypass it. This result does not establish the specified outcome for the current architecture. The decisive test must demonstrate that nAT processed-byte metrics decline while service calls succeed without internet routes. The underlying design requirement is gateway endpoints avoid per-byte NAT processing for eligible regional S3 and DynamoDB traffic.