Signature-based IaaS Performance Change Detection
Sheik Mohammad Mostakim Fattah, Athman Bouguettaya
TL;DR
The paper tackles the problem of detecting long-term performance changes in IaaS by representing service behavior as performance signatures derived from trial-user data. It introduces a novel IaaS performance noise model (spikes, attenuation, distortion) and pairs two change-detection approaches: a sliding-window, noise-agnostic method based on time-series similarity and an SNR-based method that leverages prior noise knowledge. Signature generation aggregates user experiences through privacy-preserving normalization to capture long-run QoS variability. Experimental results on synthetic and real workload data show that the SNR-based approach provides balanced detection with strong F1 scores, offering practical means to maintain accurate signatures and support long-term IaaS service selection.
Abstract
We propose a novel change detection framework to identify changes in the long-term performance behavior of an IaaS service. An IaaS service's long-term performance behavior is represented by an IaaS performance signature. The proposed framework leverages time series similarity measures and a sliding window technique to detect changes in IaaS performance signatures. We introduce a new IaaS performance noise model that enables the proposed framework to distinguish between performance noise and actual changes in performance. The proposed framework utilizes a novel Signal-to-Noise Ratio (SNR) based approach to detect changes when prior knowledge about performance noise is available. A set of experiments is conducted using real-world datasets to demonstrate the effectiveness of the proposed change detection framework.