Government incentives and unexpected repercussions (Part III)
Demand Side Reduction (DSR) is a mechanism by which large energy consumers can voluntarily reduce the amount of electricity that they use during peak times (typically between 1600hrs and 2000hrs on winter weekdays).
The UK National Grid is actively seeking companies that can commit to such reductions, on a voluntary basis, in return for a payment. The scheme will ultimately help keep energy system costs down for consumers by avoiding the need to build additional power stations. Companies who participate in the scheme not only benefit from a direct payment, but also reduce their bills because of the higher cost of energy at these times.
DSR has been successfully demonstrated in the commercial sector by companies such as Kiwi Power, which provides software that interfaces with customers' building management systems to reduce the energy used by 'non-essential' systems on demand.
A reasonable example of this might be the air conditioning systems used by a chain of hotels. Most hotel customers wouldn't notice or complain if the temperature of a hotel lobby increased or decreased by a few degrees over a short period of time. If this scenario was enacted across a modest number of hotels, using software that orchestrated the 'how and when', then a significant amount of energy could be saved without having a noticeable impact on the hotels' customers.
The potential issue with DSR in the data centre industry relates to what might be considered a 'non-essential' service, as data centres as a whole are often considered as mission-critical to the organisations that use them. DSR proponents have nonetheless mooted the possibility of reducing electricity during times of peak demand by using back-up generators or UPS systems to provide power to the data centre.
However, this simple concept, while attractive at first glance, may be fool-hardy given that those generators and UPS systems were specified so as to maintain critical services in the event of a power failure. Reducing the time period over which they provide back-up cover is presumably not an option if the systems were properly specified in the first place.
In reality, if DSR is to be of widespread use in data centres, owners will need to manage the IT services that are provided by their data centres in a much more pro-active and efficient manner. They will need to understand what IT services are running when, and be prepared to schedule such services differently while maintaining the same overall computational throughput.
For example, if historic utilisation figures show that there is significant underused computing capacity during the night, or at weekends, certain tasks (eg. monthly payroll or commercial reporting tasks) could be rescheduled from peak times to these less busy times, leaving more flexibility for (some) back-up energy systems to be run down while maintaining the required cover for critical services during peak times. Air cooling systems could also be focussed on maintaining these critical services, while other parts of the data centre could be allowed to increase in temperature.
So while DSR is both possible and potentially very worthwhile for data centres, it requires a joined-up approach to managing both facilities and IT systems, as well as a clear understanding of which end-users or applications require which resources, when. Until then, it will undoubtedly be consigned to the ‘good idea, not well-enough thought through’ category.