Resource usage modeling for runtime resource management of component-based applications

Runtime resource management is a method to enforce a predictable execution to applications with highly and transiently fluctuating resource usage. Having correct runtime predictions regarding resource usage of individual components allows making proper resource reservations, resulting in a better resource management of the componentbased applications. In this work, we present a state-based resource usage model for a component, in which states represent CPU utilization intervals. This resource model is intended to be used in a resource-aware component-based framework, applicable for expressing resource usage estimation of video components. Moreover, we introduce a simple resource prediction strategy that is using this resource model. The quality of this strategy depends on both state size and time slot length, and is given by two quality metrics introduced: the failure rate, which measures the fraction of the time slots in which the reserved budget was insufficient, and the resource waste, which measures unused budget.