Energy aware consolidation techniques for cloud computing infrastructures: state of the art and open research problems
Martedì 27 febbraio ore 11.30
Via Comelico 39 - Milano Sala riunioni al primo piano
Responsabile: Prof. Roberto Cordone
Relatore: Ing. Rolando Brondolin (Politecnico di Milano)
In the last decade cloud computing changed the shape of how data centers are built, managed and operated. From small scale to large scale data centers, cloud computing allowed to segment big physical servers into smaller units (i.e. Virtual Machines) that can be scheduled and provisioned on demand in a Infrastructure as a Service fashion. Of course, the rapid growth of the cloud computing model reduced the costs and enhanced the available computing resources, however, energy consumption and energy waste increased as well. For this reason, researchers started to investigate techniques to enable power-awareness in the data center, from power monitoring tools to optimization algorithms that relies on power consumption data and not only on performance data. Among the others, energy and power aware consolidation techniques have been successfully explored to reduce the energy waste caused by idle or under-utilised servers.
In this presentation we will take a look at the main techniques for energy efficient consolidation of VMs: at first we will look at consolidation as a particular form of bin packing problem, then we will look at some of the techniques explored in the last few years like Modified Best Fit Decreasing, Vector Bin Packing and sorting strategies, Dynamic Round Robin and Ant Colony Optimization.