Page 33 - European Energy Innovation - Summer 2017 publication

Summer 2017 European Energy Innovation 33

DATA CENTRE ENERGY EFFICIENCY

higher power efficiency: despite using
more energy, the workload capacity
of today’s models so far outstrips
that of their predecessors that the
ratio of throughput to power usage is
indisputably in their favour. Removing
the option of higher efficiency servers
would result in a net increase of server
power consumption.

The figure opposite illustrates how there will be a direct energy savings of for data centre operators will be along
unintended consequences can 2.4TWh and an indirect savings of 3.7 the same order of magnitude.
impact the sheer number of physical TWh via data centre infrastructure such
products required and the subsequent as cooling equipment. DIGITALEUROPE DIGITALEUROPE encourages the
increase in energy consumed. Only projects that the savings using server European Commission to recognise the
nine high performance servers active efficiency will result in a total risk of imposing a counterproductive
are necessary to deliver a defined power savings of 23.1TWh (14.4 efficiency measure based on a server
workload as compared to 199 lower directly by servers and 8.7 through idle power limit and work with us
performance low idling servers. This infrastructure), which is a significantly on maintaining the trend towards
representative illustration shows the higher savings potential. The CO2 delivering efficient energy consumption
negative impact on energy consumed, savings, as well as the financial savings in data centres. ●
high performance servers would use
just 12.5% of the TOTAL idle power Ratio of Data Centre Deployed Power using average of systems passingRatio of deployed power normalised to active efficiency
consumed if using the “low Idle Power” respective proposals with Active Efficiency as baseline
configuration.
150%
DIGITALEUROPE maintains that server
efficiency is best assessed through 140% 138% 133% 132%
the use of a performance based 130%
SERT weighted geomean efficiency 128%
metric, a method which determines
the overall performance per watt of 120%
servers delivering a specified workload
in a data centre. The metric was 110% 105% 106% 106% 107%
developed using an extensive dataset 100%
to assess and validate its effectiveness
and it reflects a consensus among 90%
DIGITALEUROPE’s members of how to
remove the least efficient servers from Idle 25% 50% 100%
the market.
Server Utilisation level
THE POTENTIAL SAVINGS
The Commission has projected that the ■ EU Idle Proposal ■ Modified Idle ■ Active Efficiency
energy consumption by servers will be
46TWh by 2030, and with the current
proposal including an idle limit, that

1 Malmodin, J., Lundén, D. (2016). The energy and carbon footprint of the ICT and E&M sector in Sweden 1990–2015 and beyond. Paper published and presented
at: ICT for Sustainability (ICT4S), Amsterdam, Netherlands, 30–31 August 2016. Shehabi, Arman, et al. 2016. United States Data Center Energy Usage Report. Berkeley,
CA: Lawrence Berkeley National Laboratory. LBNL-1005775. June 27.
http://www.datacenterknowledge.com/archives/2016/06/27/heres-how-much-energy-all-us-data-centers-consume/

www.europeanenergyinnovation.eu