A new logic for energy management in the data centre

In recent years, the major focus in data centre management has been running cost, with a subtext of environmental performance. In many instances it seemed that companies were focussing on facility-level management, and not really considering the role IT systems could also play. This resulted in wasted opportunities to make data centres into more economically and environmentally streamlined entities.

The underlying logic of energy consumption
In response to this, in 2007 Emerson Network Power introduced Energy Logic. Quite simply, Energy Logic it is a holistic and vendor-neutral approach to data centre efficiency. Anyone can make use of it, and any company can play a role in it. Energy Logic presented a model revealing the cascade effect that occurs when energy consumption is reduced at the component and device level, whereby energy savings are magnified throughout the facility as demand on support systems is reduced.

The concept was launched into a market which perhaps wasn’t quite ready to take its messages on board; businesses were still struggling to understand data centre energy efficiency. That was until the financial collapse which really took hold in 2008 forced most firms to tighten the corporate purse strings. This prompted an unprecedented interest in cost-effectively expanding capacity and enhancing data centre efficiency.

The 2.0 data centre
Since the inception of Energy Logic, businesses have become a lot smarter in how they manage power-hungry facilities, particularly the data centre. As a result, we decided to move the story on and we’re now discussing Energy Logic 2.0 within the industry. Interestingly, there are actually more similarities than differences between Energy Logic 2.0 and the original Energy Logic. I’m disappointed to say that to a degree, this is a reflection of the lack of progress that has been made in optimising data centre efficiency.

Today’s servers deliver more processing power with greater efficiency, resulting in higher rack densities. And yet despite this, few data centres are taking advantage of the highest efficiency components available to them. A lack of visibility into real-time data centre performance continues to limit the ability of data centre managers to leverage all of the optimisation opportunities available.

The advances in server processing power are estimated to increase the total power consumption of the 5,000 square foot (464.5 square meters) data centre established in the original Energy Logic from 1,127 kW to 1,543 kW. As in the original Energy Logic, slightly more than half of energy is consumed by IT equipment for a PUE of 1.91.

We’re all familiar with the fact that virtualisation has become a keystone of our operations. It is also the one area where the market has advanced beyond the original Energy Logic vision. In 2007, the Energy Logic base data centre assumed no virtualisation while the optimised data centre employed virtualisation on 20 percent of servers. Today, the average level of server virtualisation exceeds 30 percent. The strategies in Energy Logic 2.0 have been updated to reflect the latest technology and best practices.

Knowledge is power
The other significant change in Energy Logic 2.0 is that it takes full advantage of data centre infrastructure management (DCIM). This may seem like a relatively minor advance — updating the final strategy in Energy Logic from monitoring to DCIM — but the capabilities that DCIM enables are significant compared to monitoring, which is a component of DCIM. In fact, the visibility and control provided by DCIM is so integral to Energy Logic 2.0 that it is impossible to attribute some isolated percent of energy savings to DCIM. DCIM enables multiple Energy Logic strategies.

To summarise the overarching benefit of Energy Logic 2.0, it illustrates the on-going potential to optimise the data centre. It’s still a holistic and vendor-neutral approach, and still extols the benefits of the cascade effect.

Using a hypothetical test case, Energy Logic 2.0 shows how the energy consumption of a “typical” 5,000 square foot data centre could be cut by more than 70% using currently available technologies.
Although this figure might sound wildly ambitious, the reality is that it’s perfectly attainable. Achieving such results just requires a defined and mapped process. To this end, we’ve provided 10 checkpoints which can be used by any organisation to dramatically increase the efficiency of their data centre:
£ Low-power components: the cascade effect rewards savings at
the component level, and low-power components still offer
significant opportunities to make savings through the facility.

£ High efficiency power supplies: increasing power supply
efficiency from 86.6% to 93% (which is readily available) reduces
total data centre power consumption by 7.1%.

£ Server power management: a facility operating at just 20%
capacity may use 80% of the energy as the same facility operating
at 100% capacity. We calculate that implementing power
management can reduce total data centre energy consumption
by 10%.

£ Refreshed ICT architecture: virtualisation in the data centres
can be limited due to silos created by line-of-business
requirements. Implementing a cohesive ICT architecture in which
all data centre systems fall under the same rules and management
policies opens the door to new efficiencies.

£ Server virtualisation and consolidation: there is huge opportunity
to expand virtualised environments. Moving from 30% server
virtualisation to 60% can provide a 29% reduction in data centre
energy consumption.

£ Power architecture: from updated ECO-Mode double-conversion
UPS technology to intelligent paralleling, there are immediate
improvements to be made in power management. DCIM can
provide a holistic view of data centre energy consumption and
server utilisation, allowing organisations to optimise their
approach.

£ Temperature and airflow management: hot-aisle/cold-aisle rack
arrangement was proposed in Energy Logic, and we see
economiser modes of operation that enable temperature, humidity
and airflow management to become incredibly rich areas of
efficiency.

£ Variable-capacity cooling: cooling systems, like IT systems
themselves, need to be optimised to work more efficiently at partial
loads. This step could see organisations reducing data centre
energy consumption by an additional 2.6%.

£ High-density cooling: high-density cooling, needed to match the
demand for higher density racks, can be implemented to create an
additional 1.5% reduction in the energy consumption of the base
data centre.

£ DCIM: by collecting, consolidating and integrating data across IT
and facilities systems, data centre infrastructure management plays
a key role in this vision of the modern efficient data centre. The
real-time intelligence it provides can entirely define the optimisation
of a data centre and reap very significant savings through
unprecedented levels of insight and control.

In Energy Logic 2.0 the industry has a roadmap to a more efficient, sustainable and affordable future in the data centre industry. Best of all, it’s not contingent upon hypothetical or unproven technologies. Anyone can begin implementing this approach today, with commercially available systems, in their own data centre – and the results will speak for themselves.