Real-time Anomaly Detection & Root Cause Exploration at Uber

Summary

The What, Why and How of building an in-house anomaly detection system to manage system and business metrics monitoring to scale Uber.

Uber developed Argos - their real-time anomaly detection and root cause exploration tool in 2015 to monitor outages across millions of system and business metrics. Given Uber’s 20% M-o-M growth scenario, learn how they decided towards dynamic thresholds to maximize the impact of alerts and reduce alter fatigue caused by false alarms. Last but not least, the key goal for any anomaly detection system is to to make the output actionable - learn how Uber solved many such challenges towards building and deploying a real-time anomaly detection system for system and business monitoring.