EWMA Control Charts for Statistical Detecting QoS Violations

Abstract

A simple statistical approach based on exponentially weighted moving average (EWMA) charts for monitoring quality of service (QoS) violations is presented in this paper. The monitoring is performed by checking the actual services performance against the QoS stated in Service Level Agreement (SLA). Our approach monitors their QoS attributes and reveals early forecasts of violations in customers' demands. We verified the CUSUM and EWMA charts by utilizing several experiments and runtime data series from CLOUDSIM web services and validated our approach using a series of datasets from real-world web services.