Applicative-based automatic configuration management for virtual machines
This paper discusses experimental verification results of the tools for automatic determining optimal configuration of a virtual machine (VM) implemented based on previously developed models and methods (including the conditions of changing loads).
For information process models, machine learning algorithms with reinforcement are applied. These models are constructed automatically in typed π-calculus taking into account the VM logs.
In order to calculate the optimal configuration of the VM, a machine learning Q-algorithm has been implemented. It features reduction of terms correspondent to information processes on the basis of an abstract machine with states. The implemented method uses an applicative approach in the form of an abstract machine.