The EU has been funding a project aimed at developing ways to reduce the environmental impact of data centres. The approach is to use technologies and methodologies to measure the energy consumption of IT infrastructure in a more detailed way than before. Real-time sensing and measurement combined with intelligent processing is used for predictive energy consumption models.
Dr Massimo Bertoncini at Engineering Ingegneria Informatica in Italy headed up a team of researchers who spent 30 months on the project. It was conducted in the 'Green active management of energy in IT service centres' (GAMES) project and supported by €3m in funding from the European Commission.
The GAMES consortium focused on IT infrastructure, taking the view that any improvement in energy efficiency in the IT will automatically reduce energy consumption in the cooling/facility systems. In optimising the IT energy use, the approach takes into account the trade-off between energy-efficiency optimisation and business demands - such as Service Level Agreements (SLAs) and Quality of Service (QoS) guarantees.
According to Dr Bertoncini 'For data centres to become more efficient, it is essential to know how energy is being consumed. Our focus was therefore to develop effective monitoring solutions that allow data centre performance and processes to be adapted in real time.'
The results were tested at two large (and already relatively energy-efficient) data centres in Italy and Germany – one used for hosting legacy applications and the other for high performance computing. In both cases PUE (Power Usage Effectiveness) was reduced from around 1.35 to 1.25.
'We showed that this approach works across technologies and at different data centres designed for performing different tasks,' Dr Bertoncini said. 'There is always a trade-off between energy efficiency and performance: essentially, the more performance required, the more energy will be used. The key is finding the right balance to provide the best service at the lowest energy cost.’
The project also studied families of applications that showed similar energy-consumption behaviour patterns. It makes it possible to associate a set of best practices to achieve the best trade-off between SLAs, performance and energy consumption. The project has made these best practices available.
Apparently the team has received a lot of interest from industry and will make its solutions commercially available. It is also looking into launching a follow-up project to further improve the technology.
Review: One of the problems with the PUE metric is that it’s only a relative measure, comparing the data centre’s non-IT energy consumption with that used for the IT infrastructure. PUE can be improved by using more efficient cooling, for example, even though the IT systems themselves can still be very inefficient.
This EU-funded project looked at the more difficult part – the efficiency of the IT systems themselves. It’s based on giving data centre managers the ability to closely monitor energy use and decide the most optimum IT configuration to achieve the performance required whilst minimising the energy used.
This focus on the IT infrastructure is going to become increasingly important in the future as data centre managers look for continuing improvements in energy efficiency. What looks particularly interesting is the reference to the behaviour patterns of families of applications. Having a set of best practices for specific applications could prove a practical template for reducing data centre energy consumption.