Overview
NetOp's AI Correlation Engine is the core intelligence behind the platform. Instead of flooding operators with thousands of individual alerts, the engine groups related anomalies across devices, sites, and metrics into a single actionable incident.
How It Works
Raw Metrics (CPU, Latency, Packet Loss, QoE, VPN…)
|
v
Anomaly Detection
(static thresholds + AI baselines)
|
v
Correlation Engine
(groups related anomalies by time, topology, impact)
|
v
Incident
(root cause + insights + recommendations)
Anomaly Types
NetOp detects two classes of anomalies:
Static Threshold Anomalies Triggered when a metric crosses a user-configured threshold (e.g., CPU > 80%).
AI Baseline Anomalies The engine learns the normal behavior of each individual element over time. When a metric deviates significantly from its learned baseline — even if it doesn't cross a static threshold — an anomaly is raised.
From Anomaly to Incident
An incident is opened when the correlation engine detects that multiple anomalies share a common cause. For example:
- High latency on a WAN interface
- Degraded SLA QoE for Office365
- Elevated packet loss on the same firewall
These three anomalies are correlated into one incident: WAN Utilization Issue — rather than three separate alerts.
Incident Lifecycle
Anomalies Detected
|
v
Incident Created
|
v
Incident Updated (new observations added)
|
v
Incident Closed (when all anomalies resolve)
AI Analysis Tab
Each incident includes an Analysis tab — a natural-language chat interface. You can ask plain-English questions:
- "What caused the WAN utilization issue?"
- "How many VPN incidents occurred this week?"
- "Which site has the most critical incidents?"
The AI answers directly from the incident data, allowing conversational troubleshooting without manual investigation.
Benefits
| Traditional Alerting | NetOp AI Correlation |
|---|---|
| Hundreds of alerts per hour | One incident per root cause |
| Manual grouping required | Automatic correlation |
| No root cause | Root cause identified |
| No recommendations | Actionable recommendations |
Best Practices
- Don't disable anomalies — even low-severity ones feed the correlation engine.
- Review incidents rather than individual anomalies for most investigations.
- Use the Analysis tab to quickly understand unfamiliar incidents.
Comments
0 comments
Please sign in to leave a comment.