Last week Zen Kishimoto made a blog entry about the initiative of the Japan’s Green IT Promotion Council (JGIPC) to improve on PUE with its own metric, data center performance per energy (DPPE). The definition of the DPPE is as follows:
DPPE = ITEU * ITEE * 1/PUE * 1/ (1-GEC) with the following sub-metrics,
- ITEU = IT equipment utilization
- ITEE = IT equipment energy efficiency = Total rated capacity of IT equipment / Total rated energy consumption of IT equipment
- GEC = Green energy coefficient = Green (natural) energy / Total energy consumption of data center
As stated in the JGIPC white paper “Basically, we are designing DPPE as a metric for indicating data center productivity per unit energy. That is, this design began with DPPE = (production of data center) / (energy). When defining DPPE, it is necessary to determine the method of defining productivity, the range of energy, and other considerations. We believe that these should be linked to the effect of energy-saving measures taken by data centers. Therefore, we focused on four kinds of data center energy-saving measures to define sub-metrics.” The JGIPC have defined their own data center model that “focuses on a controllable area from a data center development and operation point of view.”
Th JGIPC energy metric takes two important things into account in comparison with the PUE metric. They look at the quality of energy or in other words how much green energy is used. There by addressing the issue of dirty sources of energy as stated in the Greenpeace report Making IT Green: “If we hope to phase out dirty sources of energy to address climate change, then – given the massive amounts of electricity needed in order to run computers, provide backup power and coordinate related cooling equipment that even energy-efficient data centres consume – the last thing we need is for more cloud infrastructure to be built-in places where it increases demand for dirty coal-fired power. The potential of ICT technologies and cloud computing to drive low-carbon economic growth underscore the importance of building cloud infrastructure in places powered by clean renewable energy.”
As Zen pointed out there is one weakness: “Although it is not possible to have all the power supplied by clean (green) power, if a data center is fueled by clean power alone, this ratio becomes infinity. Any metric that has a potential to become infinity may not be appropriate”.
The other thing that is taken into account is the issue of utilization and workload. Using energy doesn’t automatically mean that your IT equipment is busy with processing. This issue was already addressed in the McKinsey report, in conjunction with the Uptime Institute, Revolutionizing Data Center Efficiency where the CADE metric was introduced and defined as:
CADE = facility efficiency (FE) * asset efficiency (AE) where
- FE = facility energy efficiency * facility utilization
- AE = IT energy efficiency * IT utilization
Last April the 5th. the Green Grid announced that the consortium along with the U.S. Department of Energy’s Save Now and Federal Energy Management Programs, U.S. Environmental Protection Agency’s (EPA) ENERGY STAR Program, European Union Code of Conduct, Japan Ministry of Economy, Trade and Industry’s (METI) Green IT Initiative, and Japan’s Green IT Promotion Council (GIPC) have collectively reached an agreement on the guiding principles of data center energy efficiency metrics. In this initiative of Harmonizing Global Metrics for Data Center Energy Efficiency it is recommended that data centers begin to measure PUE.
That is indeed very pragmatical and you should start somewhere but a more refined metric where energy quality and utilization is also taken into account such as in DPPE and CADE is definitely needed. The third step should be that also the energy life cycle of the data center (see some of the other blog entries) should be taken in to account. And let’s be optimistic about the energy metrics harmonizing and improvement activities, the year 2010 isn’t finished yet.