Correlation
The Correlation module groups related alerts, events, and anomalies into unified incidents. Instead of responding to dozens of individual alerts during an outage, your team sees a single correlated incident with full context.
Key Features
- Rule-based correlation — Define grouping rules by service, host, time window, alert type, or custom tags
- AI-driven clustering — Automatically detect patterns and group alerts that co-occur across outages
- Topology-aware grouping — Use service dependency maps to correlate alerts across upstream and downstream services
- Deduplication — Suppress duplicate alerts within configurable time windows
- Noise reduction metrics — Dashboard showing alert-to-incident compression ratio (typical: 10:1 to 50:1)
- Custom correlation keys — Group by deployment ID, change ticket, or any custom label
How to Access
Navigate to Correlation in the left sidebar, or access correlation settings from Settings → Correlation Rules.
Basic Usage
- Navigate to Correlation → Rules
- Click New Rule to create a correlation rule
- Define the grouping criteria:
- Time window — How close in time alerts must occur (e.g., 5 minutes)
- Group by — Fields to match on (e.g., same service, same host, same alert name)
- Minimum count — How many matching alerts trigger incident creation
- Set the Output Severity for the correlated incident
- Enable the rule and monitor the Correlation Activity dashboard
- Review auto-created incidents in the Incidents module