On the use of intelligent local resource management for improved virtualized resource provision: challenges, required features, and an approach

This work aims to achieve better management of physical resources by dynamically reallocating and adjusting local resources accord-ing to demand. The research contribution should be measured through three aspects: administrative with a delegation mechanism to help in the management of Virtual Machine (VM) based large-scale systems, fine-grain allocation with an approach to improve the use of physical resources, and intelligent management with a configurable self-adapting engine aware of the application-level re-quirements. We take into account that each application has different behavior and requires different hardware and software. These dy-namic features in a cluster of VM-based resource providers present a challenge to assign properly, according to certain optimization criteria, the physical resources to the VMs. We evaluate, as a case study of resources management, the automated management of CPU in multiprocessor machines