Event-based Detection of Changes in IaaS Performance Signatures

TL;DR

This paper introduces a new event-based method for detecting changes in IaaS performance signatures by using anomaly detection and control chart analysis, validated through real-world experiments.

Contribution

It presents a novel anomaly detection technique leveraging user experience and a signature change detection method with self-adjustment for improved accuracy.

Findings

Effective detection of IaaS signature changes demonstrated

Improved accuracy with self-adjustment method

Validated with real-world datasets

Abstract

We propose a novel ECA approach to manage changes in IaaS performance signatures. The proposed approach relies on the detection of anomalous performance behavior in the context of IaaS performance signatures. A novel anomaly-based event detection technique is proposed. It utilizes the experience of free trial users to detect potential changes in IaaS performance signatures. A signature change detection technique is proposed using the cumulative sum control chart analysis. Additionally, a self-adjustment method is introduced to improve the accuracy of the proposed approach. A set of experiments based on real-world datasets are conducted to show the effectiveness of the proposed approach.