Energy- oriented Internet (Datacenters and Clouds)

Datacenters need 26 GW to work. Within datacenters, the 47% of energy consumption of ICT is for the servers and 34% is for cooling the devices. To improve the energy efficiency there are a few options in function of the situation. When there is operating, virtualization that uses virtual servers to allow resource sharing. When is idle, sleep mode.  In an intermediate situation between idle and operating, a solution could be a job aggregation.

 Another situation is the placement of datacenters. We can reduce the power to cooling the device moving the data centers to a cold places. We talks about the rumor of move the data centers to the poles.

To improve the energy awareness there are the following options: Sleep mode (implements modular architectures with hierarchical devices), elasticity capacity provisioning (adapt the capacity to the traffic fluctuations to turn off the idle servers) and the powerfarm (uses the recursive power on the procedure of a petition and allows parallel operations). The server energy model combines Sleep mode and Job aggregation to save energy, GHG and money. In multicore servers job aggregation is possible.

The developments in the areas of energy-awareness/efficiency and network/site security have been considerable but separate. However there are areas in common. A new perspective of the situation is that attacks could change in their main aims, exploiting weaknesses in power-saving and management mechanisms to disrupt services, or even attempting to increase the energy consumption of an entire farm, causing financial damages.

 It is not a priority to focus on the major power hungry device, but rather on the most energy sensible devices. The system's vulnerability to an attack would affect the energy cost, neutralize energy saving systems (attack can use just the amount necessary to avoid the triggering of the energy-saving mechanisms and increment the operating temperature. It also would exhaust the agreed power budget (exceeding contractual enforcement will result in economic penalties or even overcoming the physical power limits resulting in power outages), increment dirty emissions and leveraging upon IDS/IPS (make them consume more CPU even upon unsuccessful attempt).

In conclusion, attacks may explicitly impact energy-related issues like energy cost, energy consumption or GHC emissions. A possible solution to this problem is power capping that set a maximum power consumption threshold and operates the facility always below that value. Another solution is power monitoring system that if an increment is detected, takes the  corresponding actions to decrease the power like job de-scheduling/migrating, CPU voltage/ frequency scaling, downclocking devices or forcing sleep mode.