NEEE Conference: Data Center Efficiency and Reliability 101

Posted on July 09, 2013 by Adam Burnham

On June 20, I had the pleasure of moderating a panel at the New England Energy Efficiency Conference & Expo in Boston. The panel, “Make Your Data Center a Model of Energy Efficiency,” is close to our hearts at Markley Group. Joined by my esteemed peers and customers from the Red Sox, Schneider Electric and Electronic Environments, our panel aimed to help attendees better understand the challenges and strategies in how to make data centers as energy efficient as possible without sacrificing reliability.

As Sarah noted, the Markley Group receives a lot of interest from existing and potential customers, not only on efficient colocation strategies, but how we can help them be environmentally friendly as well. The session was packed and attendees asked many questions about which strategies save the most energy, which strategies reduce operating costs, and how to easily incorporate these strategies into their own environment.

At the Markley Group, we constantly balance our own strategy on the fine line between being as efficient as possible without removing our number one differentiator – 100% reliability at One Summer Street. If One Summer Street is not yet your solution, and you are looking to improve your existing facility, I’ve shared some tips below (I will say, however, that a purpose-built data center will make your life easier):

  • Aisle Containment Strategies – A primary strategy for organizations looking to transform their data center from an energy efficiency perspective is to contain the air in the data center. Basically, the strategy calls for isolating hot, cold or both aisles, rather than allowing hot exhaust air and cool supply air to mix. That otherwise requires ongoing and additional energy to blast cool air and control a larger volume of air that is mixed, short cycling, and presenting inefficiencies.
  • Blanking Panels – A low tech and simple solution to maintaining proper airflow in server racks, blanking panels are basically just simple pieces of plastic/metal to cover empty slots in the rack. As simple as it sounds, it is important to implement – any gaps in the cabinet can severely alter the airflow and lead to aisle mixing, which only increases overall operating costs.
  • Data Center Modeling – Those creating their own energy efficiency strategies and not relying on the expertise and experience of a colocation partner, must lean on data center modeling and mapping tools. Solutions that call upon computational fluid dynamics (CFD) to better understand airflows can help companies track rack space usage, energy costs and power usage.
  • DCiE/PUE Data Center infrastructure Efficiency (DCiE) is a metric developed by the Green Grid, measured by dividing the IT equipment power consumption by the power consumption of the entire data center. Power Usage Effectiveness (PUE) is the inverse of DCiE, dividing the entire power consumption by the IT equipment power consumption. An ideal PUE is 1.0.
  • DCIM – Data Center Infrastructure Management takes the efficiency strategy down to the server cabinet to help identify hot spots, cabinets that are over/underutilized, and better marry the infrastructure with the equipment it serves.

For any efforts and benchmarks to be meaningful and considered successful, metering and calculations should be generated regularly, at various times, to show true results. If all this sounds too complicated or scary, consider collocating, and put the process of efficiency in the hands of a trusted partner.

As Steve Conley, IT Director at the Boston Red Sox, said during our panel, “the hardest part of collocating with Markley was giving up control of the process of storing our data, but as soon as I realized that they can handle it, it made my life easier and gave me a peace of mind – I no longer have to worry about hot or cold aisle containment, or anything like that.”

Any questions? Please contact our team and we are happy to discuss further – info@markleygroup.com.