Customers Outsourcing Energy Efficiency Goals and Responsibilities to Data Center Providers

Outsourcing Power Usage Effectiveness
Outsourcing Power Usage Effectiveness

Customers under pressure to improve organizational sustainability have a solution; make your data center provider do it. More and more requests for proposal are mandating specific energy efficiency goals, usually in the form of a specific power usage effectiveness (PUE) number.

PUE is an oft-criticized metric but is easy to understand. PUE is calculated by dividing total facility energy usage by computing equipment energy usage. The lower the PUE the more energy efficient it is.  So a data center that has equal facility and IT load would have a 2.0 PUE (2 xs / x).

A typical RFP for colocation that includes a PUE target is frequently in the 1.3 range. A data center that has IT load of only 30% as much as the overall facility load would have a PUE of 1.3 (1.3x / x).

Google claims a PUE of 1.12 across the company’s data center portfolio, and they work very hard to get there. That’s excellent, but they have control of the entire footprint of their facility, the technology, the airflow, and everything else that can influence PUE.

For multi-tenant data centers things are different, and there are significant issues with mandated PUE in colocation contracts.

  1. In a facility that is new or otherwise not near capacity PUE isn’t going to be very impressive, the infrastructure that supports the entire facility still needs to be running, which makes optimizing PUE an impossible proposition until the facility is full.
  1. The customer must do their part, both in space utilization and energy efficiency. They can’t be ramping into the space over time (unused space makes attaining a low PUE difficult). The customer also must optimize their cooling (things like cold aisle containment, chimneys, and other airflow strategies), and their servers and other IT gear need to be modern and energy efficient.
  1. Other tenants in a colocation facility must be playing along as well, as they also can impact the PUE both positively and negatively for the same reasons seen above.

PUE isn’t the best metric to start with, and it really isn’t an accurate judge of a data center, especially a colocation facility, until the building is full and all tenants are committed to keeping their equipment up to date, fully utilizing their space, and optimizing the cooling within their space.

By contractually mandating a PUE goal the tenant is putting responsibility for their sustainability on a third party (the data center provider). But many factors that influence a multi-tenant data center’s PUE are out of the hands of the provider.

PUE is an imperfect metric for data centers and in a colocation environment there are a lot of moving parts that influence it. Energy efficiency in a colocation facility is a shared responsibility between data center provider and tenant. If the customer is going to make PUE demands in their contract they need to do their part.