I85PREP
⌘ K
OCI Observability Professional study materials
OCI Observability Professional study resources

OCI Observability Professional practice questions

One question at a time. Build your OCI Observability Professional understanding.

First 50 of 200 questions. Each one lists the choices, the correct answer, and an explanation. The checker at the top of this page saves progress in this browser when it is available.

Continue in the practice tool above to load more questions, 50 at a time.

Question 1Monitoring

In OCI Monitoring, which metric statistic should you choose to understand the tail latency experienced by the slowest requests over a period?

  1. Mean
  2. Count
  3. P95 percentile Correct answer.
  4. Sum

Explanation

Percentile statistics such as P95 or P99 aggregate a metric so that the reported value reflects the boundary below which that share of samples fall, which exposes tail behavior that averages hide. Mean smooths out slow outliers, Count and Sum describe volume rather than latency distribution.

Question 2Monitoring

A compute instance's CPU utilization metric stops appearing in OCI Monitoring even though the instance is running. Which action most directly restores metric ingestion for that instance?

  1. Increase the retention period of the metric namespace
  2. Create an alarm on the metric namespace
  3. Enable the monitoring plugin or agent on the instance so it can publish metrics Correct answer.
  4. Attach a new tag to the metric namespace

Explanation

Metrics for resources such as compute instances are emitted by the Oracle Cloud Agent monitoring plugin running on the instance; if that plugin is stopped or disabled, no datapoints arrive. Retention, alarms, and tags operate on metrics that already exist and cannot cause emission to resume.

Question 3Monitoring

Which OCI Monitoring construct lets you combine several metric queries into a single expression, for example dividing error count by request count?

  1. A dynamic group
  2. A service connector
  3. A metric query with a Monitoring Query Language expression Correct answer.
  4. A log group

Explanation

Monitoring Query Language expressions in a metric query allow arithmetic and functions across multiple metric streams, so ratios and derived values can be charted or alarmed on. Service connectors move data, log groups organize logs, and dynamic groups define principal membership for IAM.

Question 4Monitoring

You need an alarm to fire only when the average CPU utilization of an instance exceeds 80 percent for five consecutive minutes. Which alarm configuration achieves this?

  1. Set the notification topic to repeat every five minutes
  2. Set the trigger rule to the greater-than operator with a threshold of 80 and a pending duration of five minutes Correct answer.
  3. Set the severity to Critical and leave the pending duration at zero
  4. Set the trigger rule to the less-than operator with a threshold of 80

Explanation

The pending duration requires the threshold condition to hold across the specified window before the alarm transitions to Firing, which suppresses transient spikes. Operator and threshold define the condition, while severity and notification repetition affect routing and messaging only.

Question 5Monitoring

What is the purpose of a notification topic in an OCI Monitoring alarm?

  1. It defines the metric namespace and dimensions to query
  2. It controls the sampling interval of the emitted metric
  3. It defines the set of subscribers that receive alarm messages Correct answer.
  4. It stores the raw metric datapoints for long-term analysis

Explanation

An alarm references aNotifications topic from the Notifications service, and subscriptions on that topic determine who or what is alerted when state changes. Metric selection and interval are part of the metric query, not the topic.

Question 6Monitoring

An alarm on a metric shows state OK while the metric chart clearly exceeds the threshold. Which explanation is most likely?

  1. The alarm query uses different dimensions or a different interval than the chart Correct answer.
  2. The alarm severity is set to Info
  3. The notification topic has no confirmed subscriptions
  4. The metric namespace was recently renamed

Explanation

Alarm state is computed from the exact query stored in the alarm, so a chart built with different dimensions, statistic, or interval can legitimately disagree. Missing subscriptions or a low severity would affect delivery and labeling, not whether the threshold is judged breached.

Question 7Monitoring

Which OCI service is the native destination for routing Monitoring alarm notifications to email, PagerDuty-style webhooks, or functions?

  1. OCI Notifications Correct answer.
  2. OCI Logging
  3. OCI Vault
  4. OCI Bastion

Explanation

Alarms publish to Notifications topics, and subscriptions on those topics deliver to email, HTTPS endpoints, and Functions. Logging stores and searches logs, Vault manages secrets, and Bastion provides managed SSH access.

Question 8Monitoring

A team wants a dashboard that shows metrics from several compartments in one view. What must be true for the dashboard widgets to display those metrics?

  1. The metrics must be exported to Object Storage first
  2. All resources must reside in the tenancy's root compartment
  3. Each compartment must have its own notification topic
  4. The dashboard owner must have read access to the compartments being queried Correct answer.

Explanation

Dashboard widgets run metric queries under the viewing user's or resource principal's authorization, so read access to each queried compartment is required. Resources need not be in the root compartment, and metrics are queried directly rather than exported first.

Question 9Monitoring

Which statement correctly describes the relationship between a metric namespace and metric dimensions in OCI Monitoring?

  1. Namespaces and dimensions are both immutable once a datapoint is posted
  2. A namespace is a key-value pair, and dimensions are top-level containers of metrics
  3. Dimensions define retention, and namespaces define alarm severity
  4. A namespace groups related metrics, and dimensions are key-value pairs that identify a specific stream within it Correct answer.

Explanation

Metrics are organized under a namespace such as oci_computeagent, and each metric is further qualified by dimensions like resourceId or availability domain that identify the individual time series. Retention and alarm severity are separate configuration concerns.

Question 10Monitoring

You want to be alerted when any one of several instances in a compartment reports a disk utilization above 90 percent, without creating one alarm per instance. What should you do?

  1. Create a separate alarm for each instance and group them in a dashboard
  2. Query the metric by namespace and dimension filter for the compartment and alarm on the aggregated result Correct answer.
  3. Use a single alarm on the compartment's object count metric
  4. Rely on the default alarm that OCI creates for every instance

Explanation

A single alarm can evaluate a metric query scoped by compartment and dimension filters, so any matching resource breaching the threshold triggers it. OCI does not auto-create per-instance alarms, and object count is unrelated to disk usage.

Question 11Monitoring

Which metric statistic is most appropriate for a metric that reports the total number of bytes transferred in each interval?

  1. P50
  2. Sum Correct answer.
  3. Min
  4. Last

Explanation

For counter-style metrics where each datapoint represents an amount accumulated during the interval, Sum across the window gives total throughput. Min, median, and last sample describe distribution or a single point rather than the aggregate volume.

Question 12Monitoring

An alarm transitions to Firing and later the metric falls back below the threshold. What happens to the alarm state by default?

  1. It is deleted automatically
  2. It returns to OK after the recovery condition is satisfied Correct answer.
  3. It stays Firing permanently and a new alarm must be created
  4. It remains Firing until an operator resets it manually

Explanation

Alarms are stateful and evaluate continuously; once the metric no longer breaches the trigger rule for the required window, the state returns to OK. Manual reset, deletion, or recreation is not required.

Question 13Monitoring

You are troubleshooting why a custom metric posted through the Monitoring API is not visible in a dashboard. Which check is most relevant first?

  1. Verify the tenancy has a default notification topic
  2. Verify the dashboard has at least two widgets
  3. Verify the Logging service is enabled in the region
  4. Verify the namespace, compartment, and dimensions used in the query match those used when posting Correct answer.

Explanation

Custom metrics are stored under the namespace, compartment, and dimension set supplied at post time, and queries must match those exactly to retrieve them. Dashboard widget count, notification topics, and Logging enablement do not affect metric visibility.

Question 14Monitoring

Which OCI Monitoring feature allows you to retain and query metric data beyond the default service retention by moving it elsewhere?

  1. Increasing the alarm pending duration
  2. Exporting metrics to Object Storage or Streaming through a service connector Correct answer.
  3. Enabling the monitoring plugin on each instance
  4. Creating additional notification topics

Explanation

The Monitoring service keeps metric datapoints for a bounded retention period, so long-term retention requires exporting them to a durable store such as Object Storage or Streaming. Plugin enablement, alarm timing, and topics do not extend retention.

Question 15Monitoring

A dashboard widget shows no data for a metric that exists, and the query uses a one-minute interval over the last hour. Which cause is most plausible?

  1. The compartment has too many resources
  2. The dashboard was created in a different tenancy
  3. The metric is emitted at a coarser resolution than the requested interval Correct answer.
  4. The alarm severity is set to Warning

Explanation

If the underlying metric only publishes every five minutes, requesting a one-minute interval can yield sparse or empty results depending on how the query aggregates. Resource counts, alarm severity, and dashboard ownership do not determine datapoint availability for a valid query.

Question 16Monitoring

Which IAM policy verb is required for a user to create and manage alarms in a compartment?

  1. use
  2. manage Correct answer.
  3. inspect
  4. read

Explanation

Creating and managing alarms requires the manage verb on the alarms resource in the relevant compartment. Inspect and read are lower-privilege verbs for listing and viewing, and use applies to resources that are consumed rather than administered.

Question 17Monitoring

What distinguishes the Monitoring service from the Logging service in OCI?

  1. Monitoring stores numeric time-series metrics, while Logging stores and searches log records Correct answer.
  2. Monitoring stores log records, while Logging stores numeric time-series metrics
  3. Monitoring and Logging are the same service with different console labels
  4. Monitoring only works for compute resources, while Logging only works for databases

Explanation

Monitoring is oriented around numeric time-series datapoints that drive charts and alarms, whereas Logging ingests structured and unstructured log records for search and analysis. They are distinct services with different data models and use cases.

Question 18Monitoring

You want an alarm notification to include context about which resource breached the threshold. What should the alarm query and message configuration include?

  1. Dimension filters that identify the resource plus a message body referencing the metric and dimensions Correct answer.
  2. A severity of Info
  3. A pending duration of zero
  4. A notification topic with a single email subscriber

Explanation

Including dimensions in the query and referencing them in the alarm message body gives responders the resource identity and metric context. Subscriber count, pending duration, and severity affect delivery and urgency but not the informational content.

Question 19Monitoring

Which statement best describes how OCI Monitoring evaluates an alarm with a pending duration of ten minutes?

  1. The alarm ignores the first ten datapoints after creation
  2. The alarm samples the metric only once every ten minutes
  3. The alarm waits ten minutes after firing before sending any notification
  4. The threshold condition must be continuously satisfied across the evaluation window before the state changes Correct answer.

Explanation

Pending duration is a debounce: the trigger condition must hold throughout the window before the alarm transitions, which reduces noise from brief spikes. It does not delay notifications after firing, change sampling, or skip initial datapoints.

Question 20Monitoring

Which metric should you monitor to detect that an instance's network interface is dropping packets?

  1. CPU utilization
  2. Memory utilization
  3. Disk read bytes
  4. Network packets dropped Correct answer.

Explanation

A packets-dropped metric directly counts discarded packets on the interface and is the most direct signal of network trouble. CPU, disk, and memory metrics describe other subsystems and would at best correlate indirectly.

Question 21Monitoring

A team needs to chart a metric that is emitted only by a custom application and is not part of any OCI service namespace. What is required?

  1. Enable the compute monitoring plugin on the host
  2. Create a log group and a log
  3. Subscribe the application to a notification topic
  4. Post the datapoints to the Monitoring API under a custom namespace with appropriate dimensions Correct answer.

Explanation

Applications can publish custom metrics by calling the Monitoring API with a namespace and dimensions of their choosing, after which they can be charted and alarmed on. The compute plugin emits infrastructure metrics, log groups store logs, and notification topics deliver alerts.

Question 22Monitoring

Which combination best reduces alarm fatigue while still detecting real incidents?

  1. Tune thresholds and pending durations based on observed baselines and use severity to route appropriately Correct answer.
  2. Disable notifications and rely on dashboards only
  3. Set every alarm to the lowest possible threshold with zero pending duration
  4. Create one alarm per datapoint interval

Explanation

Effective alerting starts from observed baselines so thresholds and pending durations reflect normal variation, and severity routes urgent conditions differently from informational ones. Extremely sensitive thresholds or disabling notifications increases noise or hides incidents.

Question 23Monitoring

Which OCI Monitoring capability helps you compare a metric's current behavior against its typical behavior over the past week?

  1. Charting the metric over a longer time range with an appropriate statistic Correct answer.
  2. Creating a dynamic group for the instances
  3. Enabling the Logging service
  4. Setting the alarm severity to Critical

Explanation

Extending the time range and choosing a statistic that summarizes the distribution lets you see trends and typical values against which current behavior can be judged. Severity, dynamic groups, and Logging do not provide historical metric comparison.

Question 24Monitoring

An alarm is configured with a trigger rule using the greater-than operator, threshold 100, and interval of one minute. Which metric value causes a breach?

  1. A datapoint of 80
  2. A datapoint of 120 Correct answer.
  3. A datapoint of 0
  4. A datapoint of 100 exactly with greater-than semantics

Explanation

With a greater-than operator and threshold of 100, only values strictly above 100 breach the rule, so 120 qualifies. Values below or equal to the threshold, including exactly 100, do not satisfy a strict greater-than comparison.

Question 25Monitoring

You need to query metric data programmatically to build a custom reporting pipeline. Which interface should you use?

  1. The Bastion session API
  2. The Monitoring API's summarize metrics operation Correct answer.
  3. The Vault secret retrieval API
  4. The Object Storage PutObject API

Explanation

The Monitoring API exposes operations for querying and summarizing metric data, which is the intended path for programmatic reporting. Object Storage, Vault, and Bastion serve different purposes and do not return metric time series.

Question 26Monitoring

A Monitoring alarm uses a trigger rule with a threshold of 80 and a breach duration of 5 minutes. The metric crosses 80 and stays above it for 12 minutes. When does the alarm transition to Firing?

  1. Immediately at the first data point that exceeds 80, regardless of breach duration
  2. Only after 12 minutes, because the alarm waits for the metric to return below the threshold
  3. Never, because the breach duration must be shorter than the alarm interval
  4. After the metric has remained in breach for the full breach duration, at the end of the fifth minute of continuous breach Correct answer.

Explanation

The breach duration (pending duration) requires the metric to stay in breach continuously for that period before the alarm fires. Once five continuous minutes of breach have elapsed, the alarm moves to Firing even though the metric keeps rising afterward. The duration is measured from the start of the breach, not from when the metric returns to normal.

Question 27Monitoring

Which OCI Monitoring metric statistic should you use when you want to see the highest value a metric reached during each aggregation interval?

  1. RATE
  2. MEAN
  3. COUNT
  4. MAX Correct answer.

Explanation

MAX returns the highest data point observed within each aggregation interval, which is appropriate for peak-usage analysis. MEAN averages the points and hides spikes, while RATE and COUNT describe change and volume rather than the peak value.

Question 28Monitoring

An operations team wants a single alarm to cover every compute instance in a compartment without editing the alarm each time a new instance is launched. Which alarm configuration achieves this?

  1. Create the alarm with a resource dimension set to the OCID of the first instance
  2. Create the alarm with a query scoped to the compartment and no resource-specific dimension filter, so all matching metric streams are evaluated Correct answer.
  3. Create the alarm at the tenancy root and rely on default policies to inherit it into the compartment
  4. Create one alarm per instance and add new alarms manually as instances are created

Explanation

An alarm evaluates a Monitoring query, so scoping the query to a compartment and omitting a per-resource dimension filter lets every matching metric stream in that compartment be evaluated. New instances that emit the same metric are included automatically. Pinning a specific resource OCID or maintaining one alarm per instance defeats the goal.

Question 29Monitoring

Which OCI IAM policy statement grants a group the ability to inspect existing alarms but not create, update, or delete them?

  1. Allow group Ops to use alarms in compartment Prod
  2. Allow group Ops to inspect compartments in tenancy
  3. Allow group Ops to manage alarms in compartment Prod
  4. Allow group Ops to read alarms in compartment Prod Correct answer.

Explanation

The read verb permits listing and viewing alarms without allowing modification. The manage verb grants full control including create and delete, and use does not fit the alarm resource verb pattern here. Inspecting compartments says nothing about alarm access.

Question 30Monitoring

A team posts a custom metric with the namespace acme_apps and dimensions host and env. They then query the namespace acme_app with the same dimensions and get no data. What is the most likely cause?

  1. The namespace must match an OCI service name to be queryable
  2. The metric requires a minimum of 24 hours before it becomes queryable
  3. Custom metrics cannot be queried with dimensions attached
  4. The query references a namespace name that does not exactly match the namespace used when the metric was posted Correct answer.

Explanation

Metric queries must reference the exact namespace string used when the data was posted; acme_app and acme_apps are different namespaces. Dimensions are supported on custom metrics and do not block querying. Custom metrics are queryable shortly after ingestion, not after a day.

Question 31Monitoring

Which statement correctly describes what happens to alarm history after an alarm is deleted?

  1. The alarm definition is removed but the alarm continues to fire silently in the background
  2. The alarm definition remains but is disabled until it is recreated
  3. The alarm definition is removed, and the historical alarm state transitions associated with it are no longer available through the alarm Correct answer.
  4. The alarm definition is removed and all related metric data is also deleted

Explanation

Deleting an alarm removes its configuration, and the alarm's own history is no longer accessible through that alarm. Metric data itself is independent of alarms and is not deleted with them. A deleted alarm also cannot continue to evaluate or fire.

Question 32Monitoring

You need to visualize the same set of metrics across two dashboards that different teams maintain. What is the most maintainable approach in OCI Monitoring?

  1. Duplicate the dashboards and edit each copy independently going forward
  2. Create a separate alarm for each dashboard so the data stays in sync
  3. Export the metrics to a spreadsheet and chart them outside OCI
  4. Define the query once and reuse the same query text in widgets on both dashboards Correct answer.

Explanation

Dashboard widgets are driven by Monitoring queries, so reusing the same query text keeps the two dashboards consistent and easy to update. Independent copies drift over time, external spreadsheets lose live data, and alarms do not feed dashboards.

Question 33Monitoring

A metric is emitted every 5 minutes, but a dashboard widget queries it with a 1-minute interval. What is the most likely visual result?

  1. The chart shows the five-minute value repeated in every one-minute bucket
  2. The chart shows interpolated values that OCI computes for the missing minutes
  3. The chart shows gaps or intermittent points because no data exists for most one-minute buckets Correct answer.
  4. The chart fails to render and returns an error for the whole widget

Explanation

Aggregation buckets with no underlying data points have nothing to display, so a query interval finer than the emission frequency produces gaps. OCI Monitoring does not synthesize interpolated or repeated values for empty buckets. The widget still renders the buckets that do contain data.

Question 34Monitoring

Which OCI Monitoring concept groups related metrics under a single name so they can be queried together?

  1. Alarm
  2. Dimension
  3. Compartment
  4. Namespace Correct answer.

Explanation

A namespace is the top-level container that groups related metrics, such as those from one service or one custom application. Alarms evaluate queries, compartments organize resources, and dimensions are key-value labels that filter within a namespace.

Question 35Monitoring

An alarm notification should tell responders which compartment and metric triggered it. Which alarm setting provides this context in the notification body?

  1. The notification topic's subscription protocol
  2. The alarm interval setting
  3. The alarm severity level alone
  4. A message template that includes alarm metadata placeholders such as the metric and compartment Correct answer.

Explanation

Alarm message templates can include placeholders that render details such as the metric name, compartment, and breached value in the notification. Severity, interval, and subscription protocol control evaluation and delivery, not the descriptive content of the message.

Question 36Monitoring

Which metric statistic is best when you want to know how many data points were received for a metric in each interval, for example to detect a stopped emitter?

  1. COUNT Correct answer.
  2. MEAN
  3. RATE
  4. MAX

Explanation

COUNT reports the number of data points received in each aggregation interval, so a drop to zero reveals an emitter that stopped sending. MAX and MEAN describe value magnitude, and RATE describes change over time rather than sample volume.

Question 37Monitoring

A team wants alarms to notify a chat channel and an on-call paging tool from the same alarm. What is the correct OCI approach?

  1. Create two notification topics and reference both in a single alarm
  2. Send the alarm to email and rely on email forwarding to reach chat and paging
  3. Create one notification topic, subscribe both the chat webhook and the paging endpoint to it, and reference that topic in the alarm Correct answer.
  4. Create two alarms with identical trigger rules, each pointing to a different destination

Explanation

A notification topic supports multiple subscriptions, so one topic with both endpoints delivers the same alarm to each destination. Duplicate alarms add maintenance and can drift, an alarm references a single topic, and email forwarding is fragile and indirect.

Question 38Monitoring

Which OCI Monitoring feature lets you save a metric query so it can be run again later or reused in dashboards without retyping it?

  1. Saved queries Correct answer.
  2. Notification topics
  3. Service connectors
  4. Alarm definitions

Explanation

Saved queries store a Monitoring query for reuse in the console, API, and dashboards. Alarm definitions evaluate thresholds rather than store reusable queries, notification topics route messages, and service connectors move data to other services.

Question 39Monitoring

You are designing an alarm for a metric that is naturally noisy and briefly spikes above the threshold many times per hour. Which change most directly reduces false firing without hiding sustained problems?

  1. Add more subscribers to the notification topic
  2. Lower the threshold slightly below the typical spike value
  3. Decrease the alarm interval to one minute
  4. Increase the breach duration so only sustained breaches fire Correct answer.

Explanation

Requiring the metric to remain in breach for a longer period filters out short spikes while still catching sustained degradation. A shorter interval evaluates more often and can increase noise, lowering the threshold makes firing more likely, and adding subscribers only spreads the same notifications.

Question 40Monitoring

Which OCI Monitoring alarm state indicates that the metric has breached the threshold but the breach duration has not yet been satisfied?

  1. Firing
  2. OK Correct answer.
  3. Pending
  4. Reset

Explanation

During an initial threshold breach, an OCI Monitoring alarm remains OK until the configured trigger delay is satisfied. It then transitions to FIRING. The pending duration describes this waiting interval; Pending is not a published alarm status.

Question 41Monitoring

A custom application posts metrics with a high-cardinality dimension such as request ID. What is the most likely operational consequence?

  1. Metric streams multiply, making queries and dashboards harder to manage and potentially increasing cost Correct answer.
  2. The namespace is renamed to reflect the dimension values
  3. The dimension is silently dropped and the metric is rejected
  4. Monitoring automatically aggregates all request IDs into one stream

Explanation

Each distinct dimension combination forms its own metric stream, so high-cardinality dimensions such as request IDs create many streams that complicate querying and can raise ingestion costs. Monitoring does not silently collapse or drop such dimensions, and namespaces are not renamed.

Question 42Monitoring

Which approach best lets a team correlate an application deployment time with a change in a monitored metric on a dashboard?

  1. Annotate the dashboard timeline or compare the metric chart against deployment timestamps recorded in the same time window Correct answer.
  2. Rely on the alarm severity to indicate when the deployment happened
  3. Query the metric with a one-day interval to smooth the deployment effect
  4. Disable the alarm during deployments and re-enable it afterward

Explanation

Correlating a metric change with a deployment requires aligning the metric chart with known deployment timestamps, for example through timeline annotations. Severity does not encode deployment events, coarse intervals hide short changes, and disabling alarms removes detection rather than aiding correlation.

Question 43Monitoring

Which OCI Monitoring capability should you use to detect that a metric has stopped being emitted entirely?

  1. An alarm on the absence of data using a query that detects missing metric streams Correct answer.
  2. An alarm on the MAX statistic with a very high threshold
  3. A dashboard widget with a one-hour interval
  4. A notification topic with an email subscription

Explanation

Detecting a silent emitter requires an alarm condition based on missing data rather than on value magnitude. Thresholds on MAX only react to high values, coarse dashboard intervals are visual only, and a notification topic just delivers messages.

Question 44Monitoring

Which statement best explains why an alarm query that uses a very short interval can produce different firing behavior than the same threshold with a longer interval?

  1. The interval only affects dashboard rendering and never alarm evaluation
  2. Shorter intervals aggregate fewer data points, so brief spikes are more likely to appear in the aggregated value Correct answer.
  3. Shorter intervals always produce lower aggregated values than longer intervals
  4. Longer intervals ignore the threshold entirely

Explanation

Aggregation over a shorter interval includes fewer data points, so a brief spike contributes more to the resulting value and is more likely to breach the threshold. Longer intervals average spikes with quieter points. The interval directly affects evaluation, not just rendering.

Question 45Monitoring

Which metric statistic should you choose to see how quickly a counter is increasing per second during each interval?

  1. COUNT
  2. MAX
  3. RATE Correct answer.
  4. MEAN

Explanation

RATE computes the per-second change of a metric, which is what you want for the speed at which a counter grows. MAX and MEAN describe value magnitude, and COUNT reports the number of data points.

Question 46Monitoring

A dashboard must show metrics from a compartment that the viewing user cannot access. What is the most likely outcome?

  1. The widget shows no data because the user's IAM policies do not permit reading metrics in that compartment Correct answer.
  2. The widget shows data only if the user is a tenancy administrator
  3. The widget shows data because dashboards bypass IAM policies
  4. The widget returns an error that permanently breaks the dashboard

Explanation

Metric visibility follows IAM policy, so a user without read access to the compartment's metrics sees no data in the widget. Dashboards do not bypass authorization, and non-administrators can view data when policies grant access. The dashboard itself remains usable.

Question 47Monitoring

Which OCI Monitoring metric namespace would you query to find metrics emitted by the Object Storage service?

  1. The namespace of the compartment that holds the bucket
  2. The namespace named after the alarm that monitors the bucket
  3. A user-defined namespace created for each bucket
  4. The service-specific namespace published by Object Storage Correct answer.

Explanation

OCI services publish metrics under their own service namespaces, and Object Storage metrics are queried from that namespace. Compartments, alarms, and per-bucket user namespaces are not the source of service-published metrics.

Question 48Monitoring

A team wants to retain a copy of selected metric data for long-term trend analysis beyond what the Monitoring service keeps online. Which OCI capability should they use?

  1. Query the metric with a one-second interval to keep more data
  2. Move the metric data into Object Storage for durable long-term retention Correct answer.
  3. Increase the alarm breach duration to preserve history
  4. Create additional notification topics to archive messages

Explanation

Long-term retention of metric data is achieved by exporting it to a durable store such as Object Storage, where it can be queried later. Breach duration, notification topics, and query interval do not extend retention.

Question 49Monitoring

Which statement correctly describes the relationship between an alarm and the notification topic it references?

  1. Each subscription must reference the alarm directly to receive messages
  2. The alarm can reference only one subscription, not a topic
  3. The topic evaluates the metric and tells the alarm when to fire
  4. The alarm publishes messages to the topic, and each subscription on the topic delivers to its own endpoint Correct answer.

Explanation

An alarm publishes notifications to a topic, and subscriptions on that topic determine which endpoints receive them. The topic does not evaluate metrics, subscriptions attach to topics rather than alarms, and an alarm references a topic that can carry many subscriptions.

Question 50Monitoring

You want to compare a metric's current value against its value at the same time on previous days to spot a recurring pattern. Which query approach is most suitable?

  1. Rely on the alarm severity to reveal the recurring pattern
  2. Query only the last five minutes with a one-second interval
  3. Query a longer time range with a coarse interval and overlay or compare the same time window across days Correct answer.
  4. Query the metric with the COUNT statistic only

Explanation

Spotting recurring daily patterns requires a time range spanning multiple days so the same window can be compared across them. A very short recent window, alarm severity, or a single statistic like COUNT cannot reveal day-over-day recurrence.