How DIET is useful for Data Center management?

Do you want to learn how DIET can be used efficiently to optimize the energy consumption in Datacenter environment?

A DC Tree based on the Diane Alber's comic. You find the answer of this question during the thesis defense of Daniel Balouek-Thomert entitled « Scheduling on Clouds considering energy consumption and performance trade-offs : from modelization to industrial applications« . The defense will be held on Monday, December 5th, at 1 p.m. at École Normale Supérieure de Lyon in the Thesis Room. Slides and talk will be in English.

The massive usage of connected services in companies and homes led to a major rise of Cloud computing technologies. Cloud Computing offers an attractive economical model where the hardware running the Cloud and processing the data does not reside at a company physical site, which means that customers do not have to care about maintenance or real costs of infrastructure. Despite its financial and operational benefits, Clouds can have a negative impact on environment due to the energy consumption of Datacenters.

The Daniel Balouek-Thomert’s thesis has investigated the energy efficiency of Clouds by proposing an extensible and multi-criteria scheduling framework that intends to increase the efficiency of an heterogeneous infrastructure from the energy consumption perspective. Our approach relies on a cursor that enables the involvement of users and providers to create scheduling policies. The target is to deliver a service that is proportional to the effective demand and eliminate energy waste often due to resource over-sizing.

This research work has been validated on the Grid’5000 platform and integrated within the DIET middleware. An industrial transfer is in progress within the NUVEA platform. This platform offers support to Cloud operators and customers with services of infrastructure audit and optimization.