First, the demand for computing and storage resources, which obviously require power, is expected to increase. However, expanding the physical footprint of a data center along with new workload-accommodating power infrastructures, including transformers, power distribution units, and backup generators, is a big no-no. When IT organizations are, at best, in budgetary cost-containment, as opposed to cost-reduction, mode, obtaining the capital expense (CAPEX) dollars for a data center upgrade solely for power reasons is likely to be a very hard sell. Then there is the price of electricity, which is not likely to go down. Just the opposite. The organizations that own the operating expense (OPEX) budget are likely feeling under increasing pressure because of upward-spiraling electricity costs, but that doesn't let IT off the hook, as the enterprise still has to pay the bill for inefficient energy management.
For energy management in the data center, having good information at the granular device level is necessary as is the analytical reporting capability to make that information actionable. For example, what would be the impact of a tech refresh with more energy-efficient equipment? Or what would be the energy/cost impact of virtual server-based consolidation, which powers down no-longer-needed physical servers?
Alas, gathering good power consumption information has not been easy. Faceplate information is simply not accurate enough. For example, power utilization and consumption are dynamic, not fixed, whereas faceplates assume a fixed use of energy. How can that be? One example is that modern processing chips use more or less energy depending upon workloads.
To address this point, Viridity's EnergyCenter monitors utilization and maps that to energy consumption over time in order to create a more accurate power utilization profile. This, in turn, can be used for analysis, such as identifying underutilized IT equipment and planning when and why to make that aforementioned tech refresh.
A better way than faceplates is to use physical sensors to collect information in real time. This would make possible the use of continuous information, and it is the most accurate way of gauging energy usage. However, sensors tend to be expensive and intrusive and could cause management problems, such as trying to keep track of configuration changes and updating the sensor network, as well.