Question 110
Operations sees a rise in HTTP 429 responses despite steady end-user traffic. What metric relationship is most useful?
- Correlate throttled requests with token throughput and the deployed model quota
- Treat cost alerts as notifications and implement separate authorized enforcement if needed
- Redact or minimize sensitive prompt content and restrict access to trace data
- Review actual quota allocation by relevant model, region, subscription, and deployment
Explanation
Correct Answer
A:A request limit or token ceiling can produce throttling even if ordinary service health metrics appear normal. In the customer-facing AI service under an SLO, confirm the fix by retesting the affected behavior after correlating throttled requests with token throughput and the deployed model quota.
Incorrect Answers
B:Although useful elsewhere in the customer-facing AI service under an SLO, a standard budget threshold does not itself guarantee resource suspension or hard cost containment; the present case is different and calls for the team to correlate throttled requests with token throughput and the deployed model quota.
C:Tracing can expose the very data an AI application processes; observability needs privacy and access controls in its own workflow, but that is a different correction from the present requirement to correlate throttled requests with token throughput and the deployed model quota. The reported issue would remain.
D:That change addresses another failure mode: quota may be shared or constrained by specific dimensions; a single displayed number is not necessarily per-team capacity; it does not address this case, for which the relevant intervention is to correlate throttled requests with token throughput and the deployed model quota.
Question 111
A team responsible for a customer-facing AI service under an SLO needs to notify finance when forecast or actual consumption passes a spending threshold; the acceptance check is that a notification is raised when the cost threshold is reached. Which capability should be selected?
- Enable application tracing
- Configure Azure Cost Management budgets and alerts
- Inspect Azure Monitor service metrics
- Check quota and rate-limit configuration
Explanation
Correct Answer
B:Cost Management budgets compare scoped expenses with configured limits and can notify designated parties. For this customer-facing AI service under an SLO, the intended evidence is that a notification is raised when the cost threshold is reached. A budget alert is not a request-rate throttle or a content safety filter.
Incorrect Answers
A:Distributed traces record spans for the application workflow, revealing where time and errors occur when instrumentation is configured; it may be used to follow a request through retrieval, prompts, tools, and inference steps to isolate slow operations in a customer-facing AI service under an SLO, but the current requirement is instead to notify finance when forecast or actual consumption passes a spending threshold. That separate capability does not prove that a notification is raised when the cost threshold is reached; A trace is not an Azure billing invoice or an image classification score, so selecting that alternative would not meet the stated acceptance check.
C:Platform metrics aggregate measurable service behavior and help detect capacity or health trends; it may be used to measure resource throughput, latency, and request failures over time in a customer-facing AI service under an SLO, but the current requirement is instead to notify finance when forecast or actual consumption passes a spending threshold. That separate capability does not prove that a notification is raised when the cost threshold is reached; They do not preserve detailed prompt content or serve as financial budget limits, so selecting that alternative would not meet the stated acceptance check.
D:Quotas and rate limits constrain allowed request volume or token throughput according to the applicable service and deployment; it may be used to understand why inference traffic is throttled at peaks despite a healthy model in a customer-facing AI service under an SLO, but the current requirement is instead to notify finance when forecast or actual consumption passes a spending threshold. That separate capability does not prove that a notification is raised when the cost threshold is reached; Raising a dashboard's display interval will not increase serving entitlement, so selecting that alternative would not meet the stated acceptance check.
Question 112
A customer-facing AI service under an SLO must measure resource throughput, latency, and request failures over time. Reviewers will accept the implementation when time-series charts expose request and failure patterns. Which approach best fulfills that requirement?
- Enable application tracing
- Check quota and rate-limit configuration
- Inspect Azure Monitor service metrics
- Configure Azure Cost Management budgets and alerts
Explanation
Correct Answer
C:Platform metrics aggregate measurable service behavior and help detect capacity or health trends. For this customer-facing AI service under an SLO, the intended evidence is that time-series charts expose request and failure patterns. They do not preserve detailed prompt content or serve as financial budget limits.
Incorrect Answers
A:Distributed traces record spans for the application workflow, revealing where time and errors occur when instrumentation is configured; it may be used to follow a request through retrieval, prompts, tools, and inference steps to isolate slow operations in a customer-facing AI service under an SLO, but the current requirement is instead to measure resource throughput, latency, and request failures over time. That separate capability does not prove that time-series charts expose request and failure patterns; A trace is not an Azure billing invoice or an image classification score, so selecting that alternative would not meet the stated acceptance check.
B:Quotas and rate limits constrain allowed request volume or token throughput according to the applicable service and deployment; it may be used to understand why inference traffic is throttled at peaks despite a healthy model in a customer-facing AI service under an SLO, but the current requirement is instead to measure resource throughput, latency, and request failures over time. That separate capability does not prove that time-series charts expose request and failure patterns; Raising a dashboard's display interval will not increase serving entitlement, so selecting that alternative would not meet the stated acceptance check.
D:Cost Management budgets compare scoped expenses with configured limits and can notify designated parties; it may be used to notify finance when forecast or actual consumption passes a spending threshold in a customer-facing AI service under an SLO, but the current requirement is instead to measure resource throughput, latency, and request failures over time. That separate capability does not prove that time-series charts expose request and failure patterns; A budget alert is not a request-rate throttle or a content safety filter, so selecting that alternative would not meet the stated acceptance check.
Question 113
The solution brief for a customer-facing AI service under an SLO requires the system to route supported resource logs and metrics to a monitoring destination for investigation. A successful test confirms that analysts can query retained service events in their chosen destination. What should the team use?
- Check quota and rate-limit configuration
- Inspect Azure Monitor service metrics
- Enable Azure diagnostic settings
- Configure Azure Cost Management budgets and alerts
Explanation
Correct Answer
C:Diagnostic settings export available resource telemetry to targets such as Log Analytics or storage. For this customer-facing AI service under an SLO, the intended evidence is that analysts can query retained service events in their chosen destination. They do not automatically enforce a monthly spending ceiling.
Incorrect Answers
A:Quotas and rate limits constrain allowed request volume or token throughput according to the applicable service and deployment; it may be used to understand why inference traffic is throttled at peaks despite a healthy model in a customer-facing AI service under an SLO, but the current requirement is instead to route supported resource logs and metrics to a monitoring destination for investigation. That separate capability does not prove that analysts can query retained service events in their chosen destination; Raising a dashboard's display interval will not increase serving entitlement, so selecting that alternative would not meet the stated acceptance check.
B:Platform metrics aggregate measurable service behavior and help detect capacity or health trends; it may be used to measure resource throughput, latency, and request failures over time in a customer-facing AI service under an SLO, but the current requirement is instead to route supported resource logs and metrics to a monitoring destination for investigation. That separate capability does not prove that analysts can query retained service events in their chosen destination; They do not preserve detailed prompt content or serve as financial budget limits, so selecting that alternative would not meet the stated acceptance check.
D:Cost Management budgets compare scoped expenses with configured limits and can notify designated parties; it may be used to notify finance when forecast or actual consumption passes a spending threshold in a customer-facing AI service under an SLO, but the current requirement is instead to route supported resource logs and metrics to a monitoring destination for investigation. That separate capability does not prove that analysts can query retained service events in their chosen destination; A budget alert is not a request-rate throttle or a content safety filter, so selecting that alternative would not meet the stated acceptance check.
Question 114
Diagnostic settings were enabled but log searches return nothing for a resource. What should be checked first?
- Review actual quota allocation by relevant model, region, subscription, and deployment
- Treat cost alerts as notifications and implement separate authorized enforcement if needed
- Redact or minimize sensitive prompt content and restrict access to trace data
- Verify the selected diagnostic categories and their destination configuration
Explanation
Correct Answer
D:Creating a settings object without enabling relevant categories or a valid destination does not populate useful records. In the customer-facing AI service under an SLO, confirm the fix by retesting the affected behavior after verifying the selected diagnostic categories and their destination configuration.
Incorrect Answers
A:Although useful elsewhere in the customer-facing AI service under an SLO, quota may be shared or constrained by specific dimensions; a single displayed number is not necessarily per-team capacity; the present case is different and calls for the team to verify the selected diagnostic categories and their destination configuration.
B:That change addresses another failure mode: a standard budget threshold does not itself guarantee resource suspension or hard cost containment; it does not address this case, for which the relevant intervention is to verify the selected diagnostic categories and their destination configuration.
C:Tracing can expose the very data an AI application processes; observability needs privacy and access controls in its own workflow, but that is a different correction from the present requirement to verify the selected diagnostic categories and their destination configuration. The reported issue would remain.
Question 115
RAG responses slow down after a reranking rollout, and detailed request traces already exist. What analysis identifies the source of the regression?
- Send supported diagnostic logs to a retained Log Analytics workspace
- Handle 429 responses with bounded exponential backoff and service-aware retries
- Compare time-series request latency percentiles before and after the deployment
- Compare p95 durations of correlated retrieval and reranking spans before and after release
Explanation
Correct Answer
D:Trace percentile comparisons isolate the stage whose latency changed, whereas enabling traces a second time yields no additional diagnostic value. In the customer-facing AI service under an SLO, confirm the fix by retesting the affected behavior after comparing p95 durations of correlated retrieval and reranking spans before and after release.
Incorrect Answers
A:Diagnostic settings export supported service logs for later querying; metrics alone may not capture the operational events in its own workflow, but that is a different correction from the present requirement to compare p95 durations of correlated retrieval and reranking spans before and after release. The reported issue would remain.
B:That change addresses another failure mode: respectful retries recover from transient throttling without amplifying overload; unbounded immediate retries are counterproductive; it does not address this case, for which the relevant intervention is to compare p95 durations of correlated retrieval and reranking spans before and after release.
C:Although useful elsewhere in the customer-facing AI service under an SLO, latency distributions reveal tail delays that a simple success count or monthly bill cannot diagnose; the present case is different and calls for the team to compare p95 durations of correlated retrieval and reranking spans before and after release.
Question 116
After a release, users report slower completions but the service remains available. Which telemetry would most directly show the regression?
- Compare time-series request latency percentiles before and after the deployment
- Handle 429 responses with bounded exponential backoff and service-aware retries
- Send supported diagnostic logs to a retained Log Analytics workspace
- Create a cost budget with actual or forecast threshold notifications
Explanation
Correct Answer
A:Latency distributions reveal tail delays that a simple success count or monthly bill cannot diagnose. In the customer-facing AI service under an SLO, confirm the fix by retesting the affected behavior after comparing time-series request latency percentiles before and after the deployment.
Incorrect Answers
B:That change addresses another failure mode: respectful retries recover from transient throttling without amplifying overload; unbounded immediate retries are counterproductive; it does not address this case, for which the relevant intervention is to compare time-series request latency percentiles before and after the deployment.
C:Although useful elsewhere in the customer-facing AI service under an SLO, diagnostic settings export supported service logs for later querying; metrics alone may not capture the operational events; the present case is different and calls for the team to compare time-series request latency percentiles before and after the deployment.
D:Cost Management budgets can alert on usage thresholds; they are monitoring controls, not an automatic hard spending cap in its own workflow, but that is a different correction from the present requirement to compare time-series request latency percentiles before and after the deployment. The reported issue would remain.
Question 117
Two teams share inference quota and both assume the full quota belongs to their deployment. What should be examined?
- Correlate throttled requests with token throughput and the deployed model quota
- Treat cost alerts as notifications and implement separate authorized enforcement if needed
- Review actual quota allocation by relevant model, region, subscription, and deployment
- Redact or minimize sensitive prompt content and restrict access to trace data
Explanation
Correct Answer
C:Quota may be shared or constrained by specific dimensions; a single displayed number is not necessarily per-team capacity. In the customer-facing AI service under an SLO, confirm the fix by retesting the affected behavior after reviewing actual quota allocation by relevant model, region, subscription, and deployment.
Incorrect Answers
A:That change addresses another failure mode: a request limit or token ceiling can produce throttling even if ordinary service health metrics appear normal; it does not address this case, for which the relevant intervention is to review actual quota allocation by relevant model, region, subscription, and deployment.
B:Although useful elsewhere in the customer-facing AI service under an SLO, a standard budget threshold does not itself guarantee resource suspension or hard cost containment; the present case is different and calls for the team to review actual quota allocation by relevant model, region, subscription, and deployment.
D:Tracing can expose the very data an AI application processes; observability needs privacy and access controls in its own workflow, but that is a different correction from the present requirement to review actual quota allocation by relevant model, region, subscription, and deployment. The reported issue would remain.
Question 118
The finance team wants email before monthly Azure AI spend overruns its budget, not after the invoice. Which feature should it configure?
- Handle 429 responses with bounded exponential backoff and service-aware retries
- Send supported diagnostic logs to a retained Log Analytics workspace
- Create a cost budget with actual or forecast threshold notifications
- Compare time-series request latency percentiles before and after the deployment
Explanation
Correct Answer
C:Cost Management budgets can alert on usage thresholds; they are monitoring controls, not an automatic hard spending cap. In the customer-facing AI service under an SLO, confirm the fix by retesting the affected behavior after creating a cost budget with actual or forecast threshold notifications.
Incorrect Answers
A:That change addresses another failure mode: respectful retries recover from transient throttling without amplifying overload; unbounded immediate retries are counterproductive; it does not address this case, for which the relevant intervention is to create a cost budget with actual or forecast threshold notifications.
B:Although useful elsewhere in the customer-facing AI service under an SLO, diagnostic settings export supported service logs for later querying; metrics alone may not capture the operational events; the present case is different and calls for the team to create a cost budget with actual or forecast threshold notifications.
D:Latency distributions reveal tail delays that a simple success count or monthly bill cannot diagnose in its own workflow, but that is a different correction from the present requirement to create a cost budget with actual or forecast threshold notifications. The reported issue would remain.
Question 119
A team estimates monthly cost using only request count while token lengths vary greatly. What metering dimension is missing?
- Measure utilization against request and token rate limits before increasing load
- Inspect correlated tool-call spans and identifiers for the affected conversation
- Standardize supported diagnostic destinations and retention policies by environment
- Include model-specific input/output token use and applicable pricing meters
Explanation
Correct Answer
D:Inference billing depends on usage dimensions and model/deployment terms, not merely HTTP request volume. In the customer-facing AI service under an SLO, confirm the fix by retesting the affected behavior after including model-specific input/output token use and applicable pricing meters.
Incorrect Answers
A:That change addresses another failure mode: more concurrent calls can aggravate throttling; capacity diagnosis is necessary before scaling producers; it does not address this case, for which the relevant intervention is to include model-specific input/output token use and applicable pricing meters.
B:Request traces can reveal duplicate invocations and their order; aggregated request counts cannot prove a single workflow's actions in its own workflow, but that is a different correction from the present requirement to include model-specific input/output token use and applicable pricing meters. The reported issue would remain.
C:Although useful elsewhere in the customer-facing AI service under an SLO, consistent routing makes correlated incident analysis possible without assuming a resource automatically retains all logs forever; the present case is different and calls for the team to include model-specific input/output token use and applicable pricing meters.
Question 120
Several environments send logs to different workspaces, complicating an incident. How should ingestion be improved?
- Use aggregate metrics for trends and separate traces for individual request paths
- Measure utilization against request and token rate limits before increasing load
- Inspect correlated tool-call spans and identifiers for the affected conversation
- Standardize supported diagnostic destinations and retention policies by environment
Explanation
Correct Answer
D:Consistent routing makes correlated incident analysis possible without assuming a resource automatically retains all logs forever. In the customer-facing AI service under an SLO, confirm the fix by retesting the affected behavior after standardizing supported diagnostic destinations and retention policies by environment.
Incorrect Answers
A:Although useful elsewhere in the customer-facing AI service under an SLO, resource metrics summarize behavior across traffic; they do not reconstruct each retrieval, tool, and model operation; the present case is different and calls for the team to standardize supported diagnostic destinations and retention policies by environment.
B:That change addresses another failure mode: more concurrent calls can aggravate throttling; capacity diagnosis is necessary before scaling producers; it does not address this case, for which the relevant intervention is to standardize supported diagnostic destinations and retention policies by environment.
C:Request traces can reveal duplicate invocations and their order; aggregated request counts cannot prove a single workflow's actions in its own workflow, but that is a different correction from the present requirement to standardize supported diagnostic destinations and retention policies by environment. The reported issue would remain.