Predicting provisioning and booting times in a Metal-as-a-service system

Cloud management automation and management of SLA incidents become a research challenges for any Cloud service-based system. In the era of ongoing adoption of Cloud Computing at a fast rate the Metal-as-a-service (MaaS) platforms assure a higher level of performance, but at the cost of a more complex provisioning system, all of these being imposed by SLA assurance. More, disaster recovery and critical infrastructure protection become important aspects for any real-time applications that use Cloud Services. This paper deals with the problem of predicting provisioning and booting times in a MaaS system, and proposed a solution based on platform monitoring and a multi-variate regression algorithm. The configuration, provisioning flow, and capacity management capabilities were tested on Bigstep Full Metal Cloud platform an event-based tracking system, based on which provisioning times can be calculated for each individual element. We analyzed the performance of proposed solution by comparing the predicted booting and provisioning times with real times using different scenarios.