Time Series QoS Forecasting for Management of Cloud Services

Management of Cloud services is one of the important aspects for the cloud service users to manage in order to ensure that they achieve their required outcomes. There is a wide interest in the literature on this problem, but most of that work has approached this problem from the service provider´s (platform) viewpoint. While on the one hand, having techniques to monitor a service from this viewpoint is important, on the other hand it is also important to monitor the QoS of a cloud service being received at the user side. This is because there is a possibility of the service user being unable to obtain the promised service with the required characteristics due to factors beyond the platform side which affects the QoS being received at the run time. One of the main factors for user side service monitoring is the accurate forecasting of the QoS of cloud services over a period of time in the future based on the past observed pattern or history. In this paper we investigate the use of exponential smoothing and autoregressive moving average models for forecasting the QoS of cloud services. We propose a forecasting mechanism which uses the past QoS values collected though QoS monitoring to forecast the future QoS of cloud services.