You can download the full white paper (11 pages):
http://www.modius.com/WPF_PUE_Real_World

Excerpt from the Document:

Introduction:

As companies look for new ways to increase efficiencies and reduce
costs, attention is increasingly turning to the data center – the
capacity of which is almost universally strained by the unmitigated
expansion of IT gear, both in terms of the addition of new servers, as
well as the overall growth in the power-density of equipment.

Power Usage Effectiveness (PUE) is an emerging metric that, as its
name implies, gives data center managers and business executives a
highly useful yardstick for gauging how efficiently their data center
facility consumes power – a determination that varies significantly
depending on the compute load, the design of the cooling systems, and
general efficiency of facilities equipment.


In search of an improved PUE:

However, Gartner notes that there are some limitations with PUE as
the metric is currently used. Generally, because it is either captured
intermittently, or based on only few instrumentation points, PUE
generally only “provides a simple point-in time snapshot of the energy efficiency
of the data center.” By using PUE only in a high-level, snapshot
format, operators miss valuable insights into their data center
operations, including how:

* Efficiency varies during the course of the day
* It is influenced by seasonal and long-term business cycles
* It is correlated to other factors such as outside air temperature.

Modius has designed a solution that addresses this challenge.
Modius uniquely allows PUE to be monitored both at an extremely
comprehensive level (with as many instrumentation points as possible),
as well as continuously throughout the day. This approach captures
minute-by-minute changes in power consumption
that characterize today’s highly dynamic, globally oriented data center
environments. The company believes that continuous and comprehensive
measurement of PUE is essential to leverage this metric as a means to
pinpoint strategies that can reduce energy consumption, and ensure that
data center reliability is uncompromised.