Question 990 A high-traffic application sends enormous telemetry volume to monitoring and needs to manage ingestion while retaining useful trends. Which technique might help? Increase log-retention time as the only method of reducing ingestion volume Use an availability set to compress events Delete every error event indiscriminately without metrics Supported application telemetry sampling with awareness of analysis trade-offs Explanation Correct Answer D:Sampling can reduce volume but may omit rare events. Incorrect Answers A:Longer retention changes storage economics, not the rate at which verbose telemetry is submitted. B:Compute placement isn't telemetry sampling. C:Suppressing all failures destroys incident evidence.
Question 991 A release causes user-facing slowdowns but Azure Service Health shows no platform incident. What should the team compare? Only the subscription's billing email Application Insights performance and exception trends before and after the release Treat a healthy VM metric as proof that every user request succeeded Only VM image SKU history without request telemetry Explanation Correct Answer B:App telemetry can pinpoint code-related regressions independent of platform health. Incorrect Answers A:A contact address isn't transaction evidence. C:Host availability does not rule out failed application requests, exceptions or dependency problems. D:Infrastructure changes alone may not explain user-facing slowdown.
Question 992 A support team receives one incident reference and must locate the failing request among millions. Which telemetry field or mechanism is especially useful? A consistent operation or correlation identifier captured in traces Only the VM's resource group description Only the storage account's globally unique name Look only at a daily total of errors with no operation or dependency correlation Explanation Correct Answer A:Identifiers connect related diagnostics for the single failure. Incorrect Answers B:Administrative metadata lacks request identity. C:The account name doesn't identify one HTTP request. D:The aggregate count cannot connect an individual failed request to its underlying dependency failure.
Question 993 A mixed-language application has telemetry for its .NET API but nothing from a separate custom worker. What should developers verify? One instrumented web app guarantees traces from every external process Assume selecting an SDK package alone guarantees every background worker is instrumented Supported instrumentation or SDK/agent configuration for the worker A resource group tag automatically injects tracing into code Explanation Correct Answer C:Coverage is component-specific and not automatically inherited from another service. Incorrect Answers A:Uninstrumented dependencies may be opaque. B:Application components require appropriate instrumentation and configuration; a package name is not complete telemetry coverage. D:Metadata doesn't add SDK calls.
Question 994 An app experiences an incident during a regional Azure platform degradation. Which evidence combination is most reliable? Use only a cost forecast to diagnose unavailability Treat an Azure Service Health incident as a replacement for all application metrics Assume every Azure notice means no app-level debugging is necessary Compare relevant Service Health notices with Azure Monitor and application telemetry Explanation Correct Answer D:Provider status and measured app impact together explain the event. Incorrect Answers A:Spending doesn't confirm endpoint failure. B:Provider incidents and customer runtime behavior require different evidence to understand impact. C:Applications can have additional failures.
Question 995 A team deploys identical services in development and production but wants only production errors to page on-call engineers. What monitoring setting is relevant? Give every resource the same 'Prod' tag without changing alert scope Delete the development subscription's audit logs Apply one global alert rule without scoping it to the intended subscriptions or resources Target alert-rule scopes and filters to production resources Explanation Correct Answer D:Alert routing should represent the operational environment rather than all tests. Incorrect Answers A:Labels alone may not adjust existing rules. B:Removing logs isn't necessary to scope alerts. C:An improperly scoped rule may miss production resources or notify for unrelated environments.
Question 996 An operations group receives both informational and critical alerts and wants routing appropriate to impact. Which Azure Monitor field should it design around? Send every informational and critical alert to the same escalation destination Only the subscription creation date Only the VM OS disk SKU Alert severity and corresponding notification/action group mappings Explanation Correct Answer D:Severity helps prioritize investigation and response. Incorrect Answers A:Routing based on impact requires severity-aware actions rather than treating all events as equally urgent. B:Account age doesn't classify current incidents. C:Disk tier doesn't encode alert impact.
Question 997 A critical metric alert should invoke a preapproved remediation workflow rather than merely email an operator. What Azure Monitor integration is relevant? An action group configured for an appropriate automation endpoint A VM availability set Create an alert description with a runbook URL but no action or automation target A resource tag containing script text Explanation Correct Answer A:Actions can initiate supported webhooks or runbooks with authorization. Incorrect Answers B:Compute placement isn't an alert action handler. C:Descriptive text alone does not invoke the intended operational response when an alert fires. D:Tags don't run scripts on an alert.
Question 998 An organization wants every possible debug trace retained indefinitely but also demands minimal monitoring cost. What trade-off should be discussed? Choose appropriate ingestion, sampling, and retention by operational and compliance value Reduce query display size while keeping the same ingestion and retention Assume unlimited log ingestion and retention are free for all plans Disable all billing reports to keep telemetry cheaper Explanation Correct Answer A:Exhaustive indefinite telemetry storage can be expensive. Incorrect Answers B:Shorter query output does not by itself reduce the stored telemetry volume driving log charges. C:Azure Monitor data costs vary by configuration. D:Hiding cost views doesn't lower usage.
Question 999 A website has acceptable average CPU but sporadic external unavailability. What monitoring combination is more revealing? Availability probes and application request/failure telemetry alongside infrastructure metrics Only the subscription billing contact Only a storage account hot tier Only a larger CPU allocation without investigation Explanation Correct Answer A:CPU alone cannot establish end-user reachability. Incorrect Answers B:Billing fields aren't service health checks. C:Object temperature doesn't monitor public HTTP uptime. D:More compute may not fix DNS or application errors.
Question 1000 An Azure Monitor alert reports a failed production dependency, and the on-call engineer fixes it. What should be documented to close the incident? Only the VM's temporary disk size Evidence of recovered health from relevant metrics, logs or probes Only the engineer's verbal confidence Only a resource tag called Resolved Explanation Correct Answer B:A resolved symptom should be verified rather than assumed from the change itself. Incorrect Answers A:Disk capacity may be unrelated to dependency health. C:A change can fail or have unexpected effects. D:Metadata isn't objective recovery evidence.
