Exact forecasts for electricity markets might be difficult, but the trend is clear: energy costs are the fastest-rising expense for today’s data centers.
To curb these costs, data center and facilities teams are turning to various energy management approaches, many of which involve large upfront infrastructure investments to introduce the necessary monitoring and control capabilities. New overlay networks, however, are not the answer.
Along with driving up complexity, extra hardware increases energy consumption and lowers the net gains. Hardware solutions also introduce upgrade challenges, especially since there are multiple evolving power measurement and control protocols supported by different solution vendors.
A software alternative: Holistic energy management
Energy management software can play a key role in curbing energy costs by programmatically collecting and aggregating the real-time temperature and power consumption behaviors of data center servers. Since servers consume most of the energy in the data center, and use 60 percent of their maximum power even while idle, gaining visibility of the server operating characteristics is an important step for improving overall efficiency in the data center. Modern servers provide real-time status information, and this data can be collected over existing infrastructure by agentless, centralized energy monitoring software.
Without a solution in place to automatically collect and aggregate this data, data center managers often over-budget both power and cooling. Servers and racks are provisioned based on anticipated spikes in demand, and cooling is similarly maximized to avoid any hot spots that would otherwise damage equipment and result in service outages.
IT managers can take a more proactive stance by introducing energy management software. The solutions facilitate the adjustment of the data center energy consumption and also provide insights about operating temperature patterns. These capabilities are enabled by software solutions that provide detailed monitoring of server status. In particular, these solutions for energy management make it possible to use server inlet temperature data to create holistic views of the energy and temperature patterns. This data can be aggregated for each rack, row, or room to gain insights about energy use for various workloads and service levels.
Consider a typical use case for energy monitoring. An IT manager needs to provision a rack of servers. Without real-time monitoring software, a manager would have to rely on the manufacturer’s server power ratings, or perhaps carry out lab measurements of a fully loaded server. However, with real-time monitoring software, the data center manager can accurately determine the typical maximum power draw in a production environment. This approach has been proven to help boost rack densities by as much as 60 percent.
Dynamic controls and fine tuning
The same software approach that taps into server power and temperature status information can also introduce energy control functions. In the previous use case, for example, data center managers can drive up rack densities with confidence when they take advantage of power capping features built into today’s leading energy management software. By defining and enforcing power thresholds per rack, sudden spikes in server workloads will not result in power surges that would otherwise damage equipment.
Power capping can also help data center managers more efficiently allocate power based on service priorities or extend operation during emergencies. Equipped with accurate power characteristics and energy history, uninterrupted power supplies can be configured to extend operation times for the highest-priority servers during power outages, for example.
Thermal and hardware power consumption data can be logged and analyzed to identify trends and make appropriate adjustments to data center architectures. Temperature data can benefit in-depth airflow studies for improving cooling and airflow, and lead to more energy-efficient designs of integrated facilities systems. Many data centers today are embracing high ambient temperature operations, using insights about energy consumption and related temperature patterns to safely raise the thermostat and reduce cooling costs.
Gaining more than energy efficiency
With fine-grained real-time energy monitoring and control, architects can evolve data center designs to maximize efficiency. Also, the increased visibility and control make it possible to encourage end-user behaviors that promote energy conservation. Energy usage reports can focus attention on the major consumers, and real-time alerts can be introduced to automatically flag any behaviors that raise concerns based on the organization’s energy policies and guidelines. Alerts can also automatically trigger corrective actions, such as power capping, which ensures that operations stay within predefined thresholds.
Every data center has a unique set of priorities and policies, but energy management is going to stay at the top of the list for many years to come. The software-based energy management solutions offer practical, proven practices that boost efficiencies, protect business continuity, and promote data centers that make the best use of global energy resources.
Jeffrey Klaus is the general manager of Data Center Manager (DCM) Solutions at Intel Corporation, where he has managed various groups for over 14 years.