Analytical Modeling and Prediction of Cloud Workload

Cloud workload prediction is a very critical task for elastic scaling, because cloud manager decides what configuration sequence is to be considered for resource provisioning. Matching the demand guarantees Service Level Agreement condition (SLA) and Quality of Service (QoS) performance. Many techniques have been used in literature for workload prediction that show accepted accuracy but all of them have some limitation and challenges, like modeling architecture, training, validation and tuning. In this work we propose a subjective analytical model that studies the cloud workload, and formulates a statistical/mathematical model to describe the workload behavior. Workloads are generated from different sources, such as regular applications running on cloud, or data log from Internet of Things (IoT) systems to monitor and control physical systems. Real time cyber-physical system requires high accuracy provisioning for resources to prevent any performance degradation. Predicting and modeling different types of cloud workloads enhances system performance and increases its accuracy. This work considers cloud workload modeling with excellent prediction values and less training cost compared with other methods.