Splunksplunkanomaly-detectionalertingstatisticsproduction-ready
Anomaly Detection Queries
SPL patterns for detecting anomalous behavior including sudden traffic spikes, unusual error patterns, and baseline deviation alerts.
title: "Anomaly Detection Queries" description: "SPL patterns for detecting anomalous behavior including sudden traffic spikes, unusual error patterns, and baseline deviation alerts." category: "Splunk" language: "spl" tags: ["splunk", "anomaly-detection", "alerting", "statistics", "production-ready"] createdAt: "2025-02-05" updatedAt: "2025-02-25"
Overview
Statistical anomaly detection queries that identify unusual patterns in application traffic, error rates, and user behavior. These queries establish baselines and alert when metrics deviate significantly, enabling proactive incident response.
Code
| timechart span=5m count as requests
| eventstats avg(requests) as baseline, stdev(requests) as sigma
| eval is_anomaly=if(requests > baseline + 3*sigma, 1, 0)Notes
- Three-sigma rule: values beyond 3 standard deviations occur with only 0.3% probability normally
- Rolling window (
streamstats window=60) adapts the baseline to recent trends - Combine request rate with path diversity for bot/scanner detection
- Day-over-day comparison catches gradual degradation that absolute thresholds miss
- Set minimum thresholds (e.g.,
error_rate > 5) to avoid alerting on low-traffic noise