Free Academic Paper:

Anomaly Detection in Application Performance Monitoring Data

How Analytics and Anomaly Detection Solve the Problems Associated with Accurate Application Performance Monitoring


Performance issues and outages have significant impact on businesses due to the inadequacy of traditional monitoring methods. The complexity of underlying systems, the volume and variety of performance metrics collected and the desire to correlate unusual application logging have made it harder than ever to find problems and anomalous behavior before they impact business. Using statistical models, Bayesian methods and automated anomaly detection these problems can be solved faster and more accurately than before.

This 7 page academic paper demonstrates that:

  • The problem of monitoring effectively is well suited to analysis with a statistical model of the system state
  • Bayesian methods work especially well to formulate the system state statistical model
  • The results of applying this analytical model to data are shown to effectively pre-empt and diagnose system issues far faster and more accurately than traditional methods when operationalized with machine learning

Anomaly Detection in Application Performance Monitoring Data was written by Thomas Veasey and Stephen Dodson and a version of this paper was presented at ICMLC 2014 (6th International Conference on Machine Learning and Computing).

Download Your Copy