Log-Driven Anomaly Detection and Fault Prevention in Large-Scale SAP Landscapes

This paper presents findings from an 18-month field study across six organizations in Australia, Canada, and Denmark, examining how log pattern recognition and machine learning anomaly detection can substantially reduce the time between fault onset and detection in SAP environments. Drawing on the frameworks of Veershetty (2026) and Veershetty (2025), the study reports a 61% reduction in unplanned outages and a 54% reduction in false positive alerts.