The availability, quality, source, and cost of energy to power the data center are primary topics in our industry.
Availability in this case refers to the uptime of service from the utility provider. It’s been suggested that nearly half of all service outages can be related to power problems. Power quality is increasingly of concern because of the susceptibility of electronics to power quality degradation, especially in the context of non-traditional distribution architectures that have dropped the UPS from the solution. The source of energy is important to manage the mix of sustainable energy sources versus fossil sources. Cost, of course, is a fundamental driver of our data center energy purchase strategy.
As the subject of the Software Defined Data Center (SDDC) has emerged, a subtext addressing the issues listed above has also emerged as Software Defined Power (SDP).
Software Defined Power
At its most basic level, Software Defined Power (SDP) is a way of dynamically shifting the data processing load to locations that have more favorable utility power characteristics, given one’s own definition of favorability.
For example, if utility rates are lower at night than during the day, one might find it advantageous to move the application workload to facilities in time zones that are during nightfall. This tactic is know as “follow-the-moon” and requires one to have available facilities with very far geographic separation,… or to know the location of workload execution at your service provider.
Even with less geographic diversity, workload shifting can be predicated on levels of sustainable sourced generation capacity, or detection of power quality characteristics.
Software Defined Power is an emerging technique, and limited in applicability when multi-site models are necessary to accomplish one’s energy strategy, but for the operator who can take advantage, there are opportunities.
Do you use software-defined power?
Let us know in the comments below.