In this five-part educational series on Data Center Energy Efficiency, we have looked at the factors that impact total energy use and the opportunities for improving energy efficiency in the data center. Here are six additional factors that will impact the efficiency of your data center.
Of course, selecting a site location that is physically secure and has reliable access to power, water and communications is an important first step. Moreover with the rising awareness of “free cooling” The average and highest temperatures (as well as humidity) will directly impact the energy efficiency of the cooling systems. Of course, while not directly related the facility energy efficiency per se, attention to the cost of power and it fuel source should not be overlooked. Energy costs are highly location dependent and are based on local or purchased power generation costs (related to fuel types or sustain¬able sources such as, hydro, wind or solar), as well as any state and local taxes (or tax incentives) incentives which can provide lower costs, energy efficiency rebates or tax benefits.
Facility Occupancy and Power Loading Rate – Design vs. Actual
The energy efficiency of any data center will be directly affected by the actual percentage of the design load being used. The lower the load utilization compared to its design maximum, the lower the efficiency. This is directly related to its occupancy rate. If the site will not be heavily occupied for the first several years, a modular design should be considered to mitigate the impact of under-utilization. In addition, most data centers never operate a 100% of design load capacity primarily for ensuring equipment reliability and maintaining uptime. Depending on the organization culture, typically systems are operated at no more than 80-85% of design ratings (some may push to 90%) before it is considered “full”. This is a necessary, but prudent compromise of reliability vs. energy efficiency.
Oversize Design Capacity Impacts Efficiency
When deciding on the design capacity of the data center there are many competing factor that influence the decisions. The fear of making it too small and running out of space or power in only a few years is a very realistic scenario and fear. In recent years the growth in computing power demands and power density has made many data centers that were built less than 10 years ago functionally obsolete, a real risk of then looking for additional space or power that is not readily resolvable with a dedicated single site. Conversely, over-sizing will mitigate that risk, but decrease the energy efficiency.
One method to mitigate the potential of over or under-sized data centers is modular design. Capacity planning and modular capacity designs can help mitigate the risk of capacity or functional obsolescence. In some designs, the total space and utility capacity is designed and built upfront, but only individual sections are fully outfitted with the UPS, generators and cooling equipment. This saves both upfront capital cost and recurring maintenance expenses. Moreover, it also help improve energy efficiency at each stage, since the smaller sections are more fully occupied and operate at a higher efficiency. This modular design still allows for planned expansion, without the energy efficiency penalty of under-utilization.
Proper and Continuous Maintenance
Regardless of how well designed and built, the equipment must be maintained for proper operation. In the past, this was mostly driven by the need to ensure reliability to avoid system failure. Today, while this is still the key requirement, ensure optimum energy efficiency is also part of the maintenance goals. This is particularly true for cooling systems, whose efficiency and effectiveness falls off rapidly, if filters become clogged and cooling towers are not rigorously cleaned, as well as other required maintenance and system optimization for changing internal heat loads and external ambient weather conditions.
Power density which is reflection of how much computing equipment can be placed in each rack. A data center with a lower power density would mean that you may need to use more racks (and white space) to house the same amount of computing equipment than at a higher density site. Power density is typically expressed in two ways; watts per square foot or Kilowatts (kW) per rack, or sometimes both. This is primarily based on the design and type of the data center cooling system. Many older data centers cannot effectively or efficiently cool more than 5 kW per rack (some even less), and in some cases their energy efficiency goes down beyond 3 kW per rack. Even today, not all newer data centers can accommodate medium (5–10kW per rack) or “high-density” racks which require 10 kW or more per rack.
For the complete series on data center energy efficiency download the Data Center Knowledge Executive Guide on Data Center Energy Efficiency in a PDF format compliments of Digital Realty.