Smart Monitoring¶

Smart Monitoring uses structured detection rules to continuously analyze observability data, including metrics, logs, traces, and RUM, to identify issues such as abnormal fluctuations, performance degradation, and error spikes. Once a rule is triggered, the system generates a Smart Monitoring event and aggregates the matched rules into an analysis report, helping users centrally review anomalies, judgment basis, and change trends.

Smart Monitoring supports the following two configuration methods:

• AI-generated rules: Describe monitoring requirements in natural language, and AI generates the monitor name, detection rules, and event content.
• Built-in rules: Select a preset type based on the monitored object to directly load the corresponding fixed rules.

Whether a rule is AI-generated or built-in, the monitor executes according to the current rule snapshot after being saved. The Prompt does not participate in analysis during each detection run, and subsequent rule template updates do not automatically replace saved configurations.

Select a Monitor Type¶

Go to Monitoring > Smart Monitoring, click New Monitor, and select a type based on the monitoring target.

AI-Generated Rules¶

If you have already defined business objectives and anomaly symptoms but are not familiar with specific metrics, queries, or detection algorithms, you can choose AI Smart Monitor. After entering the Prompt, AI simultaneously generates the monitor name, one or more detection rules, and event content for you to review and adjust.

Built-in Rules¶

If the monitored object falls under a common scenario, you can directly select the corresponding type and load its built-in rules:

Create a Monitor¶

1. Obtain Detection Rules¶

• AI Smart Monitor: Enter the Prompt, then generate and review the monitor name, detection rules, and event content.
• Built-in types: View the built-in rules loaded for the current type.

For how to write, generate, and regenerate the Prompt for AI Smart Monitor, see AI Smart Monitor.

2. Adjust Detection Rules¶

Each rule displays information such as the monitored object, data scope, detection metric, calculation method, detection window, judgment method, threshold, unit, and event severity. Sudden-change rules also display the comparison window, change direction, and difference calculation method.

You can perform the following operations:

• Adjust threshold: Modify the boundary at which a rule triggers an anomaly.
• Delete rule: Move rules that are temporarily not needed to Deleted Rules so that they no longer participate in detection.
• Restore rule: Restore a rule from Deleted Rules so that it participates in detection again.

Except for the threshold, all other fields of a rule are read-only. A monitor must retain at least one active rule. When only one rule remains, it cannot be deleted.

3. Configure Events and Alerts¶

After adjusting the rules, configure the following as needed:

• Event content: Define the content displayed in events and used for alert notifications.
• Associate incidents: Automatically create or associate incidents after a rule is triggered.
• Alert policy: Configure notification rules, notification targets, and notification channels.
• Operation permissions: Specify the roles, teams, or members that can operate on the current monitor.

After completing the configuration, save the monitor. The currently active rules will participate in subsequent detection.

Manage Monitors¶

The Smart Monitoring list adds a Monitor Type column to distinguish AI Smart Monitors from different built-in types. The list retains name search, filtering, pagination, and the action menu. For new-version monitors, you can view, edit, enable, disable, or delete them according to your permissions.

View Smart Monitoring Events¶

When rules satisfy the trigger conditions, Smart Monitoring events are generated. You can click View Related Events in the monitor list, or go to Event Center > All Events and filter by df_source:smartMonitor.

Event details include event content, analysis report, extended fields, alert notifications, and history. If multiple rules are triggered in the same detection, the system aggregates the rules matched in this detection and generates one combined analysis report, rather than generating a separate report for each rule.

The analysis report displays anomaly summaries by detection dimension and provides information such as matched rules, monitoring signals, current values, judgment basis, anomaly causes, and trends. For details, see Smart Monitoring Events.

Legacy Monitors¶

Smart Monitors created before the algorithm upgrade retain their original content and execution behavior, and their configuration pages are read-only. The name, rules, thresholds, event notifications, associated incidents, alert policies, permissions, tags, and enabled/disabled status cannot be modified. To adjust the detection logic, create a new monitor.

Billing¶

Smart Monitoring counts Triggers by type:

Triggers and Credits generated by AI Smart Monitoring are counted separately. For specific rules, see Commercial Plan Billing Logic.