Monitoring Health of Large Scale Software Systems Using Drift Detection Techniques

摘要: Anomaly detection in large-scale software systems is important to guarantee smooth operation of the system. Upon an anomaly, it vital identify root cause behind anomaly decipher actionable information and prevent future incidents. Isolation causes becomes inherently difficult as number components parameters each component increase. This paper discusses successful application three drift techniques, namely meta algorithm, fixed cumulative window model Page-Hinckley test that correlate system abnormalities a large scale complex Out these, change algorithm produced best result.