Energy efficiency in Data Centers is nowadays a major challenge because of its economic and environmental impact.
The numbers regarding Data Center power consumption are outstanding:
- 1.3% of worldwide energy production in 2010
- In USA alone, Data Centers consumed 80 mill MWh/year in 2011, which means the consumption of 1,5 x NYC
- 1 data center = 25 000 houses
- More than 43 Million Tons of CO2 emissions per year (2% worldwide)
- More water consumption than many industries (paper, automotive, petrol, wood, or plastic)
The GreenLSI team is part of the Integrated Systems Laboratory (LSI) Research Group of the Electronic Engineering Dpt. at Universidad Politécnica de Madrid (Spain). Our research focuses on Energy Optimization in Data Centers.
We have developed a global strategy to allow the usage of multiple information sources to coordinate decisions to cut the total energy consumption. The optimization system uses the available knowledge about the energy demand characteristics of the applications, and characteristics of computing and cooling resources to carry out proactive optimization techniques.
Our framework takes as input all the information gathered from the Data Center. The system stores data and generates knowledge that can be used to run energy optimizations at different levels of abstraction (i.e. at the resource manager level, at the architecture, compiler, etc.). Optimizations use power models trained with real data obtained from the Data Center. Results of these optimizations are evaluated by a decision-making system, which finally proposes decisions to be executed back to the Data Center.