Large-scale computing data centers have a high cost in terms of electrical power they consume. Many ideas have been put in the work to lower and keep under control this consumption; an example is the deploying of energy efficient hardware nonetheless the overall energy consumption continues to grow due to increasing requirements from such installations: for example, in 2006 the cost of energy consumption by IT infrastructures in US was estimated as 4.5 billion dollars and it is likely to double.
Currently, data centers are estimated to consume around 11.5 billion dollars per year, therefore having an environmental impact greater than some countries such as Argentina and the Netherlands.
A software oriented approach to this problem is to leverage the widespread process usually dubbed "server consolidation": an enterprise consolidates its IT infrastructure by using virtualization techniques that allow to aggregate more than one machine on a single physical hosts. The positive aspect of virtualization is that it is a platform that can be used to develop new policies that take into account not only the reliability of the services hosted by virtual machines but also a new family of energy aware policies.
Our team has designed and prototyped a modular Open-Source tool for energy savings in virtual datacenters, named Platform-Independent Energy Savings (PIES), that allow data center administrators to achieve their energy/performance objectives through assisted user-defined policies for server consolidation. Simply put, when there is an appreciable decrease of the workload of a virtual machine, PIES automatically moves virtual machines together on a single physical machine, therefore turning it off so that energy is saved. When the workload increases, PIES automatically will balance the workload between active servers and, only if necessary, will turn on physical machines.
Peoples Links
  Ivan Di Giacomo
  Rosario De Chiara
  Vittorio Scarano