Microsoft AZ-104 Exam Dumps, Practice Test Questions

Changing aggregation may show a total count but does not separate the response-code categories. The cause remains hidden in a combined series.

A resource filter narrows which resources contribute data but does not divide one API gateway’s observations by HTTP response code. The diagnostic dimension is the discriminating signal.

A duplicate unsegmented chart still mixes failures of different response codes. It cannot reveal whether 401, 429 or 500-class responses dominate.

Maximum selects the largest sample in the bucket, not their sum. It is useful for peak analysis but not interval total.

Count of metric samples indicates how many observations contributed to a time bucket; it is not the sum of transactions those observations represent. The requested five-minute transaction total uses the supported Total aggregation.

Average divides across observations and does not express total count. It can underrepresent the number of operations across a time interval.

Removing detection avoids false positives only by hiding real anomalies. The requirement is adaptive evaluation, not silence.

An extreme static threshold could miss serious anomalies outside the weekend pattern and does not model seasonality.

Alert severity controls operational classification and routing priority, not how a static CPU threshold learns recurring weekend patterns. The same 65% comparison would still fire on expected peaks. A supported dynamic threshold uses signal history and seasonality instead.

Metric definitions are scoped; they are not globally duplicated into every service metric namespace.

Aggregation controls how samples of an already selected metric are combined. It cannot import or reveal a metric definition published under a different namespace or scope. Select the correct custom namespace and supported target resource instead.

Time granularity affects sample grouping only after a metric is selected. It does not make an unrelated custom metric visible in the current platform namespace. The picker must target the scope and namespace where it was emitted.

Explicitly selecting only dimension values visible today restricts the alert to those configured values. An API Name introduced tomorrow may not be included without a rule update. The scenario specifically requires forward coverage of newly observed API operations; the documented Select all current and future values option is designed for that need.

Removing the dimension selection produces evaluation on the aggregate metric rather than separate per-API Name time series. A large number of normal GetBlob requests could conceal abnormal DeleteBlob activity in the combined count. To keep per-API alert evaluation, retain the dimension and include future values.

Maximum selects how data points within an aggregation period contribute to the metric value. It does not select or automatically enroll new dimension labels. The question is about which API-name time series the alert monitors, not how a numeric signal is summarized within one of those series.

The two metrics measure different stages of a successful request. A large difference does not, by itself, prove a storage server bottleneck; transport or client response behavior may dominate.

Replication redundancy changes data durability and availability choices but does not remove the metric's processing-versus-end-to-end measurement boundary. The difference can persist.

Stable service processing time alone does not mean transport and client-side contributions are negligible. The higher SuccessE2ELatency specifically includes request/response time outside server-side processing; discarding that observed difference would hide a likely path for investigation.

A VM's continued existence does not imply unlimited raw platform metric history. Native Azure Monitor Metrics applies defined retention periods even while a resource remains healthy; a separate long-term destination is needed when operational trend analysis spans beyond that window.

Platform Percentage CPU is an Azure-host metric available without a guest agent, and its normal retention is not just one day. Installing AMA for guest counters would not retrospectively restore the missing platform history and isn't the stated root cause.

Activity Log stores administrative and service events, not historical platform CPU metric series. Past metric samples beyond normal retention cannot be recreated by increasing event-log retention after the fact.

Cost depends on pricing and other inputs, so it is not a direct per-minute request count derived from the chosen metric total.

Relabeling a count does not divide it by time. A total of 6,000 per hour and 6,000 per minute differ by a factor of 60.

Minimum selects the smallest reported sample and is not equivalent to dividing a cumulative count by elapsed time.

Dynamic thresholds cannot be used in a metric alert rule monitoring multiple conditions. Selecting Dynamic inside the original two-condition rule does not remove that technical restriction, even if each metric could support dynamic thresholds independently.

Workbooks help visualize and filter monitoring data. They do not automatically replace metric-alert rule evaluation or configure a learned condition for the CPU signal. Notification behavior still requires an actual supported alert rule.

Action groups determine notification and automation recipients after alert evaluation. Adding one does not change whether a metric alert rule supports dynamic thresholds across two independent conditions; the same rule-design limitation remains.

Azure Virtual Machines support multi-resource platform metric alerts. Replacing a platform metric check with separate log queries is not a universal requirement and does not identify the actual problem with the proposed cross-region scope.

Private IP addresses do not define the scope of Azure Monitor platform metric alerts. Sharing an IP would be incorrect network design and would not satisfy the regional metadata constraint for cross-resource metric alerting.

The aggregation operator controls calculation across metric samples. Changing Average to Maximum cannot change the Azure region associated with either resource, and therefore does not make a cross-region multi-resource scope valid.

Azure Monitor does not guarantee that metric sets are shared across resource providers. A missing or invalid metric can invalidate the alert design.

Azure Monitor multi-resource views do not cause one provider's metric namespace to become available on another resource type. The chart must use metrics that are actually published and supported for the target resources and namespace.

Azure resource-specific metrics are scoped by resource type and published namespace. Naming a Storage metric namespace for an App Service plan does not cause the latter to emit Storage account metrics. Inspect provider-specific metric definitions and choose a valid namespace.